Polynomials in The AIME

AIME Handout
Polynomials in the AIME
Author:
For:
naman12
AoPS
freeman66
Date:
May 13, 2020
0
−8 −6 −4 −2 0 2 4 6 8
−2
−4
−6
−8
There will be a lot of this in the handout.
I had a polynomial once. My doctor removed it. - Michael Grant, “Gone”

naman12 and freeman66 (May 13, 2020) Polynomials in the AIME
Contents
0 Acknowledgements 3
1 Introduction 4
2 Roots 4
3 Vietas 10
4 Symmetric Polynomials 13
5 Complex Numbers 16
5.1 Direct Applications to Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.2 Polar Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6 A Small Bit Of Number Theory 23
7 Advanced Algebraic Manipulations 24
8 Worked Through Problems 28
9 Parting Words and Final Problems 32
10 Hints 36
A Appendix A: Proof of Results 39
B Appendix B: Polynomial Division 40
C Appendix C: Real Roots 42
D Appendix D: Lagrange Interpolation Formula 47
E Appendix E: List of Theorems, Corollaries, and Definitions 51
2
Copyright © 2020 by Euclid’s Orchard. All rights reserved.
§0 Acknowledgements
This was made for the Art of Problem Solving Community out there! We would like to thank Evan Chen for his
evan.sty code. In addition, all problems in the handout were either copied from the Art of Problem Solving
Wiki or made by ourselves. Finally, we would like to finally thank jsharmz for countless hours of proofreading
and correcting us for all of our silly mistakes.
Art of Problem Solving Community - Specific shout out to members of Professor-Mom’s Beginner AIME Forum
and FIREDRAGONMATH16’s Intermediate Algebra Forum
Evan Chen’s Personal Sty File
Me! Say hi!
A friend of naman12’s, who spent a lot of time to make sure we weren’t making trivial mistakes.
And Evan says he would like this here for evan.sty:
Boost Software License - Version 1.0 - August 17th, 2003

Copyright (c) 2020 Evan Chen [evan at evanchen.cc]
https://web.evanchen.cc/ || github.com/vEnhance
He also helped us with the hint formatting. Evan is a LATEXgod!

And finally, please do not make any copies of this document without referencing this original one.
3
§1 Introduction
Problems in polynomials come in all different flavors. Approximately once a year (AIME I and AIME II), there is
a polynomial problem. It’s not like the problem is very trivial: here’s the fish we are trying to chase:
Problem 1 (2016 AIME I Problem 11)

Let P (x) be a nonzero polynomial such that (x−1)P (x+1) = (x+2)P (x) for every real x, and (P (2))2 = P (3).
Then P ( 72 ) = m
n , where m and n are relatively prime positive integers. Find m + n.
This is definitely polynomial problem, and trying to solve it isn’t exactly trivial, as seen after a few minutes of
attempting this. Another problem is as follows:
Problem 2 (1984 AIME Problem 15)

Determine w2 + x2 + y 2 + z 2 if
x2 y2 z2 w2
+ + + =1
22 − 1 22 − 32 22 − 52 22 − 72
x2 y2 z2 w2
+ 2 + 2 + 2 =1
42 − 1 4 − 32 4 − 52 4 − 72
x2 y2 z2 w2
+ 2 + 2 + 2 =1
62 − 1 6 − 32 6 − 52 6 − 72
x2 y2 z2 w2
+ 2 + 2 + 2 =1
82 − 1 8 − 32 8 − 52 8 − 72
Do you see a polynomial here? Well, technically yes, but not what we’ve normally seen. So how can we relate
this to what we’ve seen before? We’ll see in the following sections.
A word of advice for those who intend to follow this document: almost all problems are from the AIME; a few
HMMT and USA(J)MO problems might be scattered in, but remember we go into a fair amount of depth here.
The Appendix A: Proof of Results contains very technical results that are only included for completion, but
don’t need to be understood. Appendix B: Polynomial Division contains information on polynomial division
that could be useful. Appendix C: Real Roots talks about some strategies used to find roots (not just rational)!
Most of these start basic, but quickly jump to advanced techniques and uses.
And do you have questions, comments, concerns, issues, or suggestions? Here are two ways to contact me:
1. Send an email to realnaman12@gmail.com and I should get back to you (unless I am incorporating your
suggestion into the document, when it might take a bit more time).
2. Send a private message to naman12 by either clicking the button that says PM or by going here and
clicking New Message and typing naman12.
Please include something related to Polynomial AIME Handout in the subject line so I know what you are
talking about.
§2 Roots
We start this section with one of the most important theorems (arguably) in algebra:
4
Theorem 2.1 (Fundamental Theorem of Algebra)

Given a polynomial f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 in C[x] (polynomials with complex number
coefficients), there exists a root r ∈ C (aka f (r) = 0).
Remark 2.2. The proof is given at the end. Don’t look at it unless you believe you can handle it!
Remark 2.3. n (the highest power x is raised to in f (x)) is called the degree of f and denoted as deg f .
Exercise 2.4. Show that deg(f · g) = deg f + deg g and deg(f + g) ≤ max(deg f, deg g). This is an
important exercise. Hints: 27
Now, this looks pretty naive, but we can use polynomial division (see Appendix B: Polynomial Division for more
details) repeatedly to get the following corollary:
Corollary 2.5 (Number of Roots Corollary)

coefficients), there exists exactly n roots r1 , r2 , . . . , rn ∈ C (aka f (r1 ) = f (r2 ) = · · · = f (rn ) = 0).
Sketch of Proof. Induct on the degree of f , with n = 1 being trivial and use the Fundamental Theorem of
Algebra to reduce it to the case of n − 1.
Remark 2.6. Note that the roots need not be distinct. For example, the polynomial x2 − 2x + 1 = 0 has roots 1, 1,
which are the same.
This leads to the following corollary:
Corollary 2.7 (Zero Polynomial Corollary)

coefficients), if there are n + 1 roots, then f (x) = 0.
Now, how does this help? It’s really useful in the next problem.
Theorem 2.8 (Unique Factorization of Polynomials)

Any polynomial f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 can be expressed as f (x) = an (x − r1 )(x −
r2 ) · · · (x − rn ), where r1 , r2 , . . . , rn are the roots of f (x).
Proof. Assume that the roots of f (x) were r1 , r2 , . . . , rn (by Number of Roots Corollary), then we consider
g(x) = an (x − r1 )(x − r2 ) · · · (x − rn )
Then, consider f (x) − g(x). The terms of degree n cancel, so deg f − g is at most n − 1. However, r1 , r2 , . . . , rn
are roots of both f and g (the second because of the Zero Product Property). Thus, we get that f − g has n
roots, so by Zero Polynomial Corollary, f (x) − g(x) = 0, so f (x) = g(x).
5
Remark 2.9. For those of you who know what a UFD is, this is proving that C[x] is a UFD. We haven’t shown it is
a domain yet, but the underlying idea is this.
Thus, we get that for all roots r of f (x), x − r is a factor of f (x) by the unique factorization of polynomials.
Thus, we get this useful theorem by using the fact f (r) = 0 for all roots:
Theorem 2.10 (Factor Theorem)

x − r is a factor of f (x) if and only if f (r) = 0.
This leads to the following generalization (proved in Appendix B: Polynomial Division):
Theorem 2.11 (Remainder Theorem)

f (k) is the remainder when f (x) is divided by x − k.
The factor theorem is a special case by taking r as a root, so f (r) = 0, and so as the remainder is 0, it is divisible
by x − r. This doesn’t exactly relate to roots, but is still helpful. Let’s see an example in action:
Example 2.12 (AMC 12A 2017/23)

For certain real numbers a, b, and c, the polynomial
g(x) = x3 + ax2 + x + 10
has three distinct roots, and each root of g(x) is also a root of the polynomial
f (x) = x4 + x3 + bx2 + 100x + c.
What is f (1)?
Solution. Let’s see how to approach this. We get by the Factor Theorem, g(x) = (x − a)(x − b)(x − c) (as there
can not be any more roots by Zero Polynomial Corollary), and f (x) is divisible by x − a, x − b, x − c. Thus, f (x)
is divisible by their product, and in particular,
f (x) = g(x)h(x)
for some other polynomial h(x). So now, we need to find f (1), which seems pretty daunting. However, we can
try to extract more information from h(x). What’s the degree? Well, we have that if the degree of h(x) is d,
then g(x) has 3 roots and h(x) has d roots, so then f (x) has 3 + d roots. But we know f (x) has degree 4, so
3 + d = 4! Thus, we can write h(x) = x + r for some unknown root. Then, we get that
x4 + x3 + bx2 + 100x + c = (x3 + ax2 + x + 10)(x + r)
Thus, we get that

x4 + x3 + bx2 + 100x + c − (x3 + ax2 + x + 10)(x + r) = 0 (*)
Looking at the coefficient of x3 , we get
1 − (a + r) = 0
or a + r = 1. Now, looking at the coefficient of x in the expansion of (*), we get
100 − (10 + r) = 0
6
so r = 90. Then, we get that a = −89. Now, we know g(x) = x3 − 89x2 + x + 10. Do we use that to find roots?
Well, we have no good way. Let’s try our idea of computing coefficients in (*); the coefficient of x2 is
b − (ar + 1) = 0
We know ar + 1...but isn’t that a bit big?
Remark 2.13. When the problem gets too big, try to take a step back and look at it from a different perspective.
Most of the time, if you have some substansial progress, start to work backwards.
We need to find f (1). That’s the same as g(1)h(1). Do we know g(1)? Well, we get
g(1) = 1 + a + 1 + 10 = 1 − 89 + 1 + 10 = −77,
h(1) = 1 + r = 1 + 90 = 91,
so
f (1) = g(1)h(1) = −77 · 91 = −7007 .
Remark 2.14. A question that may arise: how did we know that h(x) was in the form of x + r and not 2(x + r)?
The answer is once again to use (*) and the fact that if the coefficient of x4 (leading term) was k, then 1 − k = 0, so
k = 1. This is a small detaIl I glossed over; make sure you understand why this is valid.
Remark 2.15. So what if we didn’t notice that f (1) = g(1)h(1)? We can still find b and c, right? We can compute
b = ar + 1 = −89 · 90 + 1 = −8009
and similarly c = 10r = 900. Thus, we can plug it in to get
f (1) = 1 + 1 − 8009 + 100 + 900 = −7007
as well.
So our first remark showed us an important point - polynomials can have the exact same set of roots because
of the leading coefficient can vary while the roots are the same - something I talked above in the first remark.
For example, we can take f (x) = (x − 1)(x − 2) and g(x) = 2(x − 1)(x − 2), which are obviously not the same
polynomial, but the roots of both are 1, 2. Make sure you don’t forget this!
Our first problem – solved. And don’t think it’s easy - it is one of the harder problems on an AMC 12. I’ll leave
you with a few exercises:
7
Exercise 2.16 (2018 AIME I Problem 1). Let S be the number of ordered pairs of integers (a, b) with
1 ≤ a ≤ 100 and b ≥ 0 such that the polynomial x2 + ax + b can be factored into the product of two (not
necessarily distinct) linear factors with integer coefficients. Find the remainder when S is divided by 1000.
Hints: 71
Exercise 2.17 (2007 AIME I Problem 8). The polynomial P (x) is cubic. What is the largest value of k for
which the polynomials Q1 (x) = x2 + (k − 29)x − k and Q2 (x) = 2x2 + (2k − 43)x + k are both factors of
P (x)? Hints: 57
Exercise 2.18. Let N be the number of complex numbersa z with the properties that |z| = 1 and z 6! − z 5!
is a real number. Find the remainder when N is divided by 1000. Hints: 50
a
See Complex Numbers for more information. The only pieces of information you will need: z is a root if and only if z is
needed, and for zz = |z|2 .
Now, let’s look at a theorem that is very useful when dealing with polynomials:
Theorem 2.19 (Rational Root Theorem)

Given a polynomial P (x) = an xn + an−1 xn−1 + . . . + a1 x + a0 with integral coefficients, an =
6 0. The
a p
Rational Root Theorem states that if P (x) has a rational root r = ± q with p, q relatively prime positive
integers, p is a divisor of a0 and q is a divisor of an .
a p
A number that can be expressed as q
for integers p and q. Typically, we deal with rational numbers in lowest terms, which
just means gcd(p, q) = 1.
Sketch of Proof. Plug it in and multiply by q n . Rearranging and factoring gives −a0 q n = p(a1 q n−1 + a2 pq n−2 +
· · · + an pn−1 ). The left hand side is an integer and so is the second part of the right hand side, so we get p | a0 q n .
Because p and q are relatively prime, p | a0 . A similar result follows by isolating the term −an pn for q | an .
But what does this mean? Well, we can find all rational roots as follows1 : take the leading coefficient (say a)
and take the constant term (or the last term) (say b). Then, we get that if pq is a root, p has to divide b and
q has to divide a. This helps a lot as it gives an algorithm to give a (possibly long) list of solutions (that are
rational). Let’s look at this example:
Example 2.20 (naman12)

Find all rational roots of 6x3 + x2 − 19x + 6.
Solution. [Walkthrough] I’ll only explain a walkthrough on how to solve this:
(a) Write all factors of our leading coefficient (positive and negative).
(b) Write all factors of our constant term.
(c) Consider the following: do we need to write all negative factors for part (b)?
(d) Consider all roots. Try to find them.
1
Keep a watch; not all roots are rational - they can also be irrational or complex. In addition, not all of these roots are definitely
roots of the polynomial - they are all possible roots.
8
Doing the rational root theorem is pretty tedious, so we can use this with polynomial division in Appendix B:
Polynomial Division For now, let us proceed with the following AIME problem:
Example 2.21 (AIME I 2011/9)

Suppose x is in the interval [0, π/2] and log24 sin x (24 cos x) = 23 . Find 24 cot2 x.
Solution. At first look, where are the polynomials? We’ll need to manipulate our original condition. We can
exponentiate to get
24 cos x = (24 sin x)3/2
Squaring, we get
(24 sin x)3 = (24 cos x)2
Now what? Well, we know from trigonometry2 sin2 x + cos2 x = 1, so we can substitute to get
p
(24 1 − sin2 x)2 = (24 sin x)3
so expanding and dividing by 242 , we get
1 − sin2 x = 24 sin3 x
Now, take y = sin x, then we get
1 − y 2 = 24y 3
Rearranging, we get
24y 3 + y 2 − 1 = 0
Here’s where the Rational Root Theorem comes in handy: we use it and check that y = 13 is a root. Now, this
means that y − 31 is a factor. However, this is not very nice - however we can multiply by 3 (see my earlier
remark on leading coefficients) to rid of fraction to get 3y − 1 divides our previous polynomial, so sin x satisfies
(3y − 1)(8y 2 + 3y + 1) = 24y 3 + y 2 − 1 = 0
But the quadratic factor has no real roots3 ! So y = sin x = 13 . And then, we get because cos2 x + sin2 x = 1
√ √
(the standard Pythagorean identity) and cos x > 0, cos x = 2 3 2 . This gives cot x = cos
sin
x
x = 8, so our answer is
√ 2
24 cot2 x = 24 · 8 = 24 · 8 = 192 .
Remark 2.22. Even if a problem doesn’t have a polynomial, it’s possible that it was intended to be a polynomial
problem.
Let’s cap it off with a few exercises:
Exercise 2.23 (2018 AIME II Problem 6). A real number a is chosen randomly and uniformly from the
interval [−20, 18]. The probability that the roots of the polynomial x4 + 2ax3 + (2a − 2)x2 + (−4a + 3)x − 2
m
are all real can be written in the form , where m and n are relatively prime positive integers. Find m + n.
n
Hints: 47
Exercise 2.24 (1988 Canadian Mathematical Olympiad Problem 1). For what real values of k do 1988x2 +
kx + 8891 and 8891x2 + kx + 1988 have a common zero? Hints: 35
2
See the trigonometry handout for more information.
3
This can be checked by taking the discriminant of the quadratic.
√ In general, for a quadratic ax2 + bx + c, if it has (any) real roots,
we can use the quadratic formula to get that the term b − 4ac has to indeed be a real number, so b2 − 4ac ≥ 0 and b2 ≥ 4ac.
2
In this problem, taking a = 8, b = 3, and c = 1. 32 = 9 ≤ 32 = 4 · 8, so there are no real roots.

9
§3 Vietas
Vieta’s Formulas are a set of formulas that relate the roots of a polynomial to the coefficients of the polynomial.
Call the symmetric sum of the numbers k1 , k2 , . . . , km taking p at a time as4
X
σp = ka1 ka2 · · · kap
1≤a1 <a2 <...<ap ≤m
Isn’t this complicated?
Remark 3.1. Don’t let fancy notation scare you! It normally is easier if you break the given equation down.
a1 , a2 , . . . , ar , what do they mean? Well, as this is a handout, I have5 to include the formal definition of everything
- which can get a bit tedious, and a bit of an eyesore. However, we can take a look at this.
Example 3.2
Find the symmetric sum of just one number k taking one at a time.
Solution. Isn’t this just k? Think about it. If we are choosing numbers 1 ≤ a1 ≤ 1, we have to have a1 = 1.
Easily, this means that σ1 = k.
Looking at the footnote above, there isn’t much else to do for one number. Let’s take two numbers:
Example 3.3
Find the symmetric sum of just two numbers k1 , k2 taking one and two at a time.
Solution. Well, let’s first tackle taking them one at a time. Well, we get that 1 ≤ a1 ≤ 2, so we have two choices:
a1 = 1 or a2 = 2. So σ1 = k1 + k2 .
Now, if we’re taking two at a time, we get that 1 ≤ a1 < a2 ≤ 2, so a1 = 1 and a2 = 2. Thus, σ2 = k1 k2 .
Exercise 3.4. Can you find what the symmetric sum of three numbers k1 , k2 , k3 are taking one, two, and
three at a time? Hints: 67
Exercise 3.5. Try expanding (x − k1 ), (x − k1 )(x − k2 ), and (x − k1 )(x − k2 )(x − k3 ). Do the results seem
familiar? Hints: 26
Exercise 3.6. Can you generalize what σn is for n variables? Hints: 61
Now, we look at the following theorem that showcases the main idea of this section.
4
Technically, we define alternative values for p > m and p ≤ 0. See Newton’s Formulas (just the statement of the theorem) for more
details.
5
Well, not have, but should, at some point.
10
Theorem 3.7 (Vieta’s Formulas)

Suppose that the roots to f (x) = cn xn + cn−1 xn−1 + · · · + c0 are r1 , r2 , . . . , rn . Then, we get that if σk is
the symmetric sum taking the ri k at a time, then
cn−k
σk = (−1)k
cn
Sketch of Proof. Just take the expansion of f (x) as cn (x − r1 )(x − r2 ) · · · (x − rn ). We choose cn so that the
leading coefficients of f (x) (in the expanded form and factored form) match. To find σk , look at the coefficient
of xn−k . On one hand, in the expanded form, it’s obviously cn−k . However, now consider how many x’s you will
need from (x − r1 )(x − r2 ) · · · (x − rn ) in the factored form. Then, see how many −rj ’s you can choose, as these
will be from the factors we didn’t choose an x from. Show that this sum is just (−1)k c0 σk .
Now, how do these help? Let’s look at an application:

1
2001
Find the sum of the roots, real and non-real, of the equation x2001 + 2 −x = 0, given that there are
no multiple roots.
Solution. Well, let’s see what happens if we try to pseudo-expand it. But first we need the following well-known
theorem:
Theorem 3.9 (Binomial Theorem)

n n n−1
n n
(x + y) = x + x y + ··· + xy n−1 + y n
1 n−1
Equipped with this, we get that

2001 2
2001 1 2001 2001 2001 1 2000 2001 1
x + −x =x −x + · x − x1999 + · · ·
2 1 2 2 2
What’s important are the first and second nonzero terms - from these, we can use Vieta’s (k = 1, which is also
the sum of the roots) to get the sum of the roots is
1 2

2001
2 2 2000
= = 500 .
2001
1 4
1 ·
2
That’s a simple answer. Is there another way to get this answer? Here is a hint: note that if x is a root, so is
1
2 − x. I leave the rest as an exercise:
Exercise 3.10 (AIME I 2001/3). Solve AIME I 2001/3 with the above method. Hints: 16
Exercise 3.11. Prove Binomial Theorem. Hints: 13
Exercise 3.12 (2014 AIME I Problem 5). Real numbers r and s are roots of p(x) = x3 + ax + b, and r + 4
and s − 3 are roots of q(x) = x3 + ax + b + 240. Find the sum of all possible values of |b|. Hints: 31
11
Let’s see another example:
Example 3.13 (AIME 1996/5)

Suppose that the roots of x3 + 3x2 + 4x − 11 = 0 are a, b, and c, and that the roots of x3 + rx2 + sx + t = 0
are a + b, b + c, and c + a. Find t.
Solution. Now, at first, this looks daunting. But let’s write down what we know and what we want to find. We
have by Vieta’s that
a + b + c = −3
ab + bc + ac = 4
abc = 11
and we want to find
t = −(a + b)(b + c)(c + a) = −(2abc + a2 b + ab2 + b2 c + bc2 + c2 a + ca2 )
That doesn’t look too nice, right? In the next section, we’ll see how to deal with this. However, we can try
something else. Let’s look at each of our terms in (a + b)(b + c)(c + a). We know a + b + c = −3, so a + b = −3 − c.
That’s pretty nice. We can then rewrite t (after cancelling out negative signs) as:
t = (3 + c)(3 + b)(3 + a)
which expands to
t = 27 + 9(a + b + c) + 3(ab + bc + ac) + abc = 23 .
Now, that looks a lot like something we’ve seen before - 27, 9, 3, and 1. So let’s see if there is a shorter way to
get this solution. We get that
t = −(−3 − c)(−3 − b)(−3 − a)
Let’s replace −3 with k, to make it look more symmetric. We get
t = −(k − a)(k − b)(k − c)
Wait. By Factor Theorem, we have k 3 + 3k 2 + 4k − 11 = f (k) = (k − a)(k − b)(k − c). That’s interesting. We get
t = −f (k) = −f (−3) = 23 .
So there does exist a nice solution! This shows that there typically is a nice solution to most AIME Prob-
lems.
Exercise 3.14 (2005 AIME I Problem 8). The equation 2333x−2 + 2111x+2 = 2222x+1 + 1 has three real
roots. Given that their sum is m
n where m and n are relatively prime positive integers, find m + n. Hints: 44
Exercise 3.15 (1993 AIME Problem 5). Let P0 (x) = x3 + 313x2 − 77x − 8 . For integers n ≥ 1 , define
Pn (x) = Pn−1 (x − n) . What is the coefficient of x in P20 (x) ? Hints: 60
Exercise 3.16 (2008 AIME II Problem 7). Let r, s, and t be the three roots of the equation
8x3 + 1001x + 2008 = 0.
Find (r + s)3 + (s + t)3 + (t + r)3 . Hints: 17
12
§4 Symmetric Polynomials
We actually are going to go expand something from last section. Remember Vieta’s Formulas? Turns out, we
are going to use the same notation for σk .
Definition 4.1 (Elementary Symmetric Polynomial) — An elementary symmetric polynomial is any

multivariate (in more than one variable, like x1 , x2 , . . .) polynomial defined as taking the sum of x1 , x2 , . . . , xn
k at a time - basically σk .
Furthermore, we have the following definition
Definition 4.2 (k-Variable Symmetric Polynomial) — A symmetric polynomial in k variables is basically

a polynomial when switching any two of the variables leaves the polynomial unchanged. For example, in
x + y + z − xyz, switching any two of x, y, z don’t change the polynomial. However, in x + y + z − x2 z,
switching x and y changes the polynomial to x + y + z − y 2 z.
Now, this leads to the following powerful theorem:
Theorem 4.3 (Fundamental Theorem of Symmetric Polynomials)

Any symmetric polynomial can be expressed as the sum/product of multiple (not necessarily different)
symmetric polynomials.
For example, try this exercise:
Exercise 4.4. Show that x21 + x22 + · · · + x2n = σ12 − 2σ2 . Hints: 40
Now, the proof is once again given in the appendix. But this doesn’t tell us much - as an analogy, the theorem
tells us there is another planet outside of Earth, but not how to find it6 , where to find it, and anything about it.
Now, that does make sense. There are basically infinitely many symmetric polynomials - we can’t have all of them.
But, there are quite a few that appear very frequently, and these are given in the following relation:
Theorem 4.5 (Newton’s Formulas)

Let ρk be xk1 + xk2 + · · · + xkn . Then, we get
k
X
kσk + (−1)j σk−j ρj = 0
j=1
where we define for j > n and j < 0 σj = 0 and for j = 0 σj = 1.
Proof. We can write this as the sum/product of a bunch of symmetric polynomials as guaranteed by the
Fundamental Theorem of Symmetric Polynomials. Suppose that x1 , x2 , . . . , xn were the roots of P (x). We use
Vieta’s Formulas to get
k
(−1)j σk−j xji
X
P (x) =
k=0
Now, We have the following claim:
6
The clever Mr. Quines pointed out that the proof does give a greedy algorithm to find it, so more fun to the reader!
13
Claim 4.6 — Newton’s Formulas is true for n = k.
Proof. Just add up all terms in the form

n
(−1)j σk−j xji = 0
X
P (xi ) =
j=0
which finishes off the proof.
For k > n, the result follows by considering the polynomial
P (x)xn−k
and applying Claim 2.5 (as the zeroes contribute nothing). Now, to prove the other side, we just consider
g(x) = an xk + an−1 xk−1 + · · · + an−k
where P (x) = an xn + an−1 xn−1 + · · · + a0 . From this, we can use the Claim on g(x).
This is the slickest proof I actually know. It’s pretty short. Let’s see if you got a hang of this:
Exercise 4.7. Find ρ1 , ρ2 , ρ3 , only in terms of σ1 , σ2 , σ3 (no ρ1 , ρ2 , ρ3 ). Compare your answer for ρ2 to
your answer to Exercise 4.4. Hints: 43
Where do these come up in the AIME? Correction: where do these not come up in the AIME? Let’s look at the
following example:

Suppose that the sum of the squares of two complex numbers x and y is 7 and the sum of the cubes is 10.
What is the largest real value that x + y can have?
Solution. Now, we will use the result of Exercise 4.7. We get that
7 = ρ2 = σ12 − 2σ2
10 = ρ3 = σ13 − 3σ1 σ2 + 3σ3 = σ13 − 3σ1 σ2

Oh look! It wants us to find σ1 ! We can get from the first equation
σ12 − 7
σ2 =
2
so substituting this into the second equation, we get
2
−σ13 + 21σ12

3 σ1 − 7
10 = σ1 − 3σ1 =
2 2
so σ13 − 21σ12 + 20 = 0. Using the Rational Root Theorem, we get that the solutions are σ1 = 1, 4, −5, so the
maximum value is 4 .
14
Exercise 4.9. Find the values of σ2 for each value of AIME 1983/5. Are you glad we didn’t find the values
of x and y? Hints: 39
Exercise 4.10 (2019 AIME I Problem 8). Let x be a real number such that sin10 x + cos10 x = 11 36 . Then
sin12 x + cos12 x = m
n where m and n are relatively prime positive integers. Find m + n. Hints: 80
Let’s take a look at another example:
Example 4.11 (AIME II 2003/9)

Consider the polynomials P (x) = x6 − x5 − x3 − x2 − x and Q(x) = x4 − x3 − x2 − 1. Given that z1 , z2 , z3 ,
and z4 are the roots of Q(x) = 0, find P (z1 ) + P (z2 ) + P (z3 ) + P (z4 ).
Solution. Although this is Newton’s Sums, we don’t bash. So, we note that we need to find
ρ6 − ρ5 − ρ3 − ρ2 − ρ1
Now, we can use our knowledge of Newton’s Sums. We get from Newton’s Formulas,
ρ6 − ρ5 − ρ4 − ρ2 = 0
so thus, we can reduce what we want to find (by subtracting the equations) to
ρ4 − ρ3 − ρ1
Similarly from Newton’s Formulas, we get
ρ4 − ρ3 − ρ2 − 4 = 0
so we need to find
ρ2 − ρ1 + 4
Now, we note that
ρ1 = σ1 = 1
and
ρ2 = σ12 − 2σ2 = 12 − 2(−1) = 3
Thus, our desired answer is
ρ2 − ρ1 + 4 = 6
So don’t rip in blindly, make manipulations and then finish off the problem with little computation.
Exercise 4.12 (2015 AIME II Problem 14). Let x and y be real numbers satisfying x4 y 5 + y 4 x5 = 810 and
x3 y 6 + y 3 x6 = 945. Evaluate 2x3 + (xy)3 + 2y 3 . Hints: 10
Exercise 4.13 (1973 USAMO Problem 4). Determine all the roots, real or complex, of the system of
simultaneous equations
x+y+z =3
x2 + y 2 + z 2 = 3
x3 + y 3 + z 3 = 3
Hints: 1
15
I know the last one is technically a USAMO problem, but it’s easier than the other AIME Problem. Finally, I
would like to write down four commonly seen factorizations (last one due to AoPS user dchenmathcounts, and is
known as the Sophie Germain Identity):
(a + b + c)(ab + bc + ac) − abc = (a + b)(b + c)(a + c)
(a + b + c)(a2 + b2 + c2 − ab − bc − ac) = a3 + b3 + c3 − 3abc

(b2 + b + 1)(b2 − b + 1) = b4 + b2 + 1
a4 + 4b4 = (a2 + 2ab + 2b2 )(a2 − 2ab + 2b2 )
§5 Complex Numbers
So, what happens if we try to do √
−1
We shall define the two solutions to this ±i. We also note that7
√
x = −1 ⇐⇒ x2 = −1 ⇐⇒ x2 + 1 = 0
so thus we can “factor” by using difference of squares:
x2 + 1 = (x − i)(x + i)
i is called the imaginary unit. What happens if we add a real number and an imaginary unit (like 5i)? Well,
this gets to
z = a + bi
But what do we call it? We call it the following:
Definition 5.1 (Complex Number) — A complex number z = a + bi (for real a and b) is the sum of a
real number and imaginary number.
Note that all real numbers and pure imaginary numbers are also complex. Now, what happens when we add?
We consider
z = a + bi, w = c + di
Then, we get
z + w = (a + bi) + (c + di) = a + (c + di) + bi = (a + c) + (di + bi) = (a + c) + (b + d)i
where we used the associativity of addition (we add the imaginary and real numbers separately). What about
multiplication? It’s slightly different (using the distributive property):
(a + bi)(c + di) = a(c + di) + bi(c + di) = ac + adi + bci + bdi2 = ac + adi + bci − bd = (ac − bd) + (ad + bc)i
Note we brought our answer in the form x + yi, which is pretty normal to do. Similarly, we can define division
and subtraction. Now, we come to one of the most important definitions:
Definition 5.2 (Conjugate) — The conjugate of the complex number z = a + bi is denoted as z and has
value a − bi.
Let’s try the following:

7
⇐⇒ means that the first statement is true if and only if the second statement is true.
16
Example 5.3
Suppose z = a + bi. Find zz.
Solution. We have that

(a + bi)(a − bi) = a2 + abi − abi − b2 i2 = a2 + b2
Note this gives a very easy way to divide. We get that as
(a + bi)(a − bi) = (a + bi)(a + bi) = a2 + b2
we have
1 a − bi
= 2
a + bi a + b2
Also try to verify the following:
Exercise 5.4 (Conjugate Addition). z + w = z + w Hints: 62
Exercise 5.5 (Conjugate Multiplication). z · w = z · w Hints: 28
Exercise 5.6. Prove that z n = z n . Hints: 54se Conjugate Multiplication (a.k.a the previous exercise).
Exercise 5.7. z = z. Hints: 33
Exercise 5.8. f (z) = z. Hints: 48
Now, what happens when we imagine plotting complex numbers on a plane? We can do that and indeed define
this plane as the Argand Plane, while the one in which we plot (x, y) is the Cartesian Plane. These planes are
essentially identical except for one key caveat - the Cartesian Plane plots x versus y while the Argand Plane
plots Re(z) versus Im(z). We can define the following:
Definition 5.9 (Modulus/Magnitude) — The modulus or magnitude of a complex number z is denoted

as |z| and is the distance from z to the origin, which is 0, in the complex plane.
√
Remark 5.10. This can be seen (by the Pythagorean Theorem) as a2 + b2 , where z can be put in the Cartesian
Plane as (a, b).
We thus see that by our last example, we get that zz = |z|2 .
Exercise 5.11. Show that |z||w| = |zw|. Hints: 34
Exercise 5.12 (Real Number Conjugate). Consider a real number r. Then r = r. Hints: 20
§5.1 Direct Applications to Polynomials

We have the following beautiful result:
17
Theorem 5.13 (Complex Conjugate Theorem)

z is a root of a polynomial with real coefficients if and only if z is.
Proof. Take a polynomial P (x) = an xn + an−1 xn−1 + · · · + a0 such that z is a root. Then, we get this means
0 = P (z) = an z n + an−1 z n−1 + · · · + a0
Now, what can we do? Well, one thing we can do is to take the conjugate. We get
0 = 0 = an z n + an−1 z n−1 + · · · + a0
Now, we can use Conjugate Addition to get
0 = an z n + an−1 z n−1 + · · · + a0
Now, we can use Conjugate Multiplication to get
0 = an (z)n + an−1 (z)n−1 + · · · + a0
Finally, as a0 , a1 , . . . , an are real numbers, we can use Real Number Conjugate to get
0 = an (z)n + an−1 (z)n−1 + · · · + a0 = P (z)
so z is a root. The “if” part can be resolved as z = z.
Let’s see an application:

For certain real values of a, b, c, and d, the equation x4 + ax3 + bx2 + cx + d = 0 has four non-real
√ roots. The
product of two of these roots is 13 + i and the sum of the other two roots is 3 + 4i, where i = −1. Find b.
Solution. We call the roots w = p + qi, w = p − qi, z = r + si, z = r − si.
Remark 5.15. z and w are commonly used for complex numbers (as opposed to x and y, which are for real variables).
We note that
w + w = 2p
is a real number, and so is z + z. Thus, we get that either w + z or w + z = 3 + 4i. By symmetry, it doesn’t
matter, so we assume w + z = 3 + 4i. Then, we get
w + z = w + z = 3 − 4i.
In addition, we get
w · z = 13 + i,
so thus it’s conjugate is
wz = w · z = 13 − i.
Now, let’s look at what we want to find. By Vieta’s, we know that
b = zw + zz + zw + wz + ww + wz = 26 + (zz + zw + wz + ww)
But how to finish? We know z + w, so let’s see is we can factor the last term. Aha!
zz + zw + wz + ww = (z + w)(z + w) = (3 + 4i)(3 − 4i) = |3 + 4i|2 = 32 + 42 = 25
So we finish by getting
b = 26 + 25 = 51 .
So the Complex Conjugate Theorem is actually pretty helpful.

18
Exercise 5.16 (2013 AIME I Problem 13). There are nonzero integers a, b, r, and s such that the complex
number r + si is a zero of the polynomial P (x) = x3 − ax2 + bx − 65. For each possible combination of a
and b, let pa,b be the sum of the zeros of P (x). Find the sum of the pa,b ’s for all possible combinations of a
and b. Hints: 14
Exercise 5.17 (2013 AIME II Problem 12). Let S be the set of all polynomials of the form z 3 + az 2 + bz + c,
where a, b, and c are integers. Find the number of polynomials in S such that each of its roots z satisfies
either |z| = 20 or |z| = 13. Hints: 51
§5.2 Polar Complex Numbers

Imagine standing at the origin of the Cartesian Plane. If I gave you a direction and asked you to walk 5 steps, it
might be difficult to write the point that I end up at without Trigonometry. Instead, we have made a system to
do exactly this! It is called the Polar Coordinate system. Polar numbers (technically coordinates) are numbers
in the form of (r, θ), where r is a real number and θ is an angle (in radians). Typically, we try to use polar
coordinates as an alternative to the standard rectangular coordinates (on the cartesian plane), and we can do
the same thing here.
105◦ 90◦ 75◦

120◦ 60◦
135◦ 45◦
150◦ 30◦
165◦ 15◦
180◦ 0◦
1 2 3 4 5 6 7
195◦ 345◦
210◦ 330◦
225◦ 315◦
240◦ 300◦
255◦ 270◦ 285◦
Figure 1: The Polar Coordinate Plane.
We note that we can scale a complex number to another complex number with magnitude 1 (by dividing by
|z|). Then, we get that if
z = a + bi, a2 + b2 = 1,
we can substitute (a, b) = (cos θ, sin θ). Thus,
z z
= = cos θ + i sin θ
|z| r
so we can scale up to get
z = r(cos θ + i sin θ)
19
This is the polar form of a complex numbers (while previously we had the rectangular form a + bi). From this
eloquent form, we have the following formula:
Theorem 5.18 (De Moivre’s Theorem)
cos nθ + i sin nθ = (cos θ + i sin θ)n
Sketch of Proof. We can induct on n. When n = 1, the result is immediate. Otherwise, we have
cos nθ + i sin nθ = cos((n − 1)θ + θ) + i sin((n − 1)θ + θ)
Using our addition formulas8 , we get cos nθ + i sin nθ is equal to
cos((n − 1)θ) cos θ − sin(n − 1)θ sin θ + i(sin((n − 1)θ) cos θ + sin θ cos((n − 1)θ))
However, we also have
(cos θ + i sin θ)n = (cos θ + i sin θ)n−1 (cos θ + i sin θ) = (cos(n − 1)θ + i sin(n − 1)θ)(cos θ + i sin θ)
Expanding and simplifying should give these are equal.
Euler used this theorem to prove the following result:
Theorem 5.19 (Euler’s Formula)
eiθ = cos θ + i sin θ
This is considered one of the most beautiful results in mathematics, especially when θ = π.
Remark 5.20. Actually, we “define” i = (1, π2 ) (as if we were plotting on a complex plane), but this is completely
arbitrary, as we could have chosen i = (1, 3π
2 ). This doesn’t matter too much, but it is pretty important in the terms
of complex numbers. It simplifies numbers a lot.
This theorem helps a lot. We can take another look at the proof of De Moivre’s Theorem using Euler’s
Formula:
Theorem 5.21 (De Moivre’s Theorem)
cos nθ + i sin nθ = (cos θ + i sin θ)n
Sketch of Proof. We use Euler’s formula to get

n
cos nθ + i sin nθ = einθ = eiθ = (cos θ + i sin θ)n
from Euler’s Formula.
Now, try the following exercise:
8
See the trigonometry handout.
20
Exercise 5.22. What is e2πi ? What about eπi ? Hints: 78
Exercise 5.23. What is eiθ eiγ ? Hints: 22
Now, let’s consider the polynomial z n = 1, for some complex number z. First, let’s find |z|. We get that |z|n = 1,
so |z| = 1 (as |z| is a distance, so it is nonnegative and real). Now, we can write
z = cos θ + i sin θ (β)
We consider θ in the form (for integer k)

2kπ
θ= (Γ)
n
By De Moivre’s Theorem, we get that
z n = (cos θ + i sin θ)n = cos nθ + i sin nθ = cos 2πk + sin 2πk = 1
Thus, this means that as there as n distinct values of θ for k = 0, 1, . . . , n − 1 (note that beyond this it repeats),
we get by the Fundamental Theorem of Algebra that these are the only (distinct, remember that we can keep
looping around the circle!) solutions. These are called the nth roots of unity:
Definition 5.24 (Roots of Unity) — The nth roots of unity are roots of z n = 1.
Exercise 5.25. Find the third and fourth roots of unity. Hints: 72
Exercise 5.26. Show that 31 | 531 + 517 + 1 (or 31 divides 531 + 517 + 1). 79
We can plot all of these roots of unity and also all other points z with |z| = 1. We get something like the
following graph:
Imaginary
θ
Real
21
The roots of unity actually do something more than expected - they form a regular polygon:
Imaginary
Real
That’s an example with n = 5, or the fifth roots of unity!
Remark 5.27. Figuring out why exactly this is a regular pentagon is simple - because the idea is easy to see, I will
leave this as an exercise. Think about the angles between the roots of unity - more accurately, how are they spaced?
Now, let’s try the following problem:

Let z1 , z2 , z3 , . . . , z12 be the 12 zeroes of the polynomial z 12 − 236 . For each j, let wj be one of zj or izj .
12
X √
Then the maximum possible value of the real part of wj can be written as m + n where m and n are
j=1
positive integers. Find m + n.
Solution. We note that it’s obvious that these (the zj ’s) aren’t roots of unity - we have to divide z by 236 to get
one. Let 8xk = zk for each k; we thus get that
x12
k − 1 = 0,
so thus
2πk 2πk πk πk
xk = cos + i sin = cos + i sin
12 12 6 6
π
Now, we note that multiplying by i adds to the imaginary part, so thus we get that
2
(3 + k)π (3 + k)π
ixk = cos + i sin
6 6
We don’t care about the imaginary part, so let’s drop it. Then, we get (by√manually plugging in the numbers - I
won’t do it here) that the sum of our desired xk has maximum value 2 + 2 3. However, we still have to multiply
by 8 to get our answer as
√ √ √
m + n = 16 + 16 3 = 16 + 768 =⇒ m + n = 16 + 768 = 784 .
22
Why won’t I do it? It turns out this is manual computation checking if

(3 + k)π πk
cos ≥
6 6
and plugging in 12 values isn’t exactly my strongest suit. You can check them if you want.
Exercise 5.29. Check that indeed what I claimed is correct.
Exercise 5.30 (2019 AIME II Problem √ 8). The polynomial f (z) = az 2018 + bz 2017 + cz 2016 has real
√
coefficients not exceeding 2019, and f 1+2 3i = 2015 + 2019 3i. Find the remainder when f (1) is divided
by 1000. Hints: 24
Exercise 5.31 (1996 AIME Problem 11). Let P be the product of the roots of z 6 + z 4 + z 3 + z 2 + 1 = 0
that have a positive imaginary part, and suppose that P = r(cos θ◦ + i sin θ◦ ), where 0 < r and 0 ≤ θ < 360.
Find θ. Hints: 69
§6 A Small Bit Of Number Theory

The core of this section is the following result:
Theorem 6.1 (Difference of Polynomials)

Let P (x) be a polynomial with integer coefficients. Then, we have that a − b | P (a) − P (b).
Sketch of Proof. Try to expand P (x) = cn xn + cn−1 xn1 + · · · + c0 . Then, substitute a and b, and use the
factorization:
k−1
!
X
ak − bk = (a − b) ai bk−i = (a − b)(ak−1 + ak−2 b + · · · + abk−2 + bk−1 )
i=0
Looking at Difference of Polynomials, we get that it isn’t too deep - how will this help us? Well, let’s take a look
at this problem:

Let P (x) be a polynomial with integer coefficients that satisfies P (17) = 10 and P (24) = 17. Given that
P (n) = n + 3 has two distinct integer solutions n1 and n2 , find the product n1 · n2 .
Remark 6.3. One non-rigorous way to prove this is that we take P (x) as a quadratic. It isn’t 100% rigorous though.
So, this is the number theory section. Let’s try to see what we know - we do know that
P (n1 ) = n1 + 3
We’ll only focus on n1 , get two solutions, and then use one as n1 and the other as n2 . Well, the only thing I can
think of to induce Difference of Polynomials is to take the difference:
n1 − k | P (n1 ) − P (k) = n1 + 3 − P (k)
23
We only know 3 values of P (k) (that are not n1 ) - k = n2 , 17, 24. We can plug them all in
n1 − n2 | n1 + 3 − (n2 + 3) = n1 − n2
n1 − 17 | n1 + 3 − 10 = n1 − 7
n1 − 24 | n1 + 3 − 17 = n1 − 14
The first is useless. The second and third give the numbers
n1 − 7 n1 − 14
,
n1 − 17 n1 − 24
are integers. There’s an n1 in the numerator and denominator in each of them, so we can rewrite the fractions as
n1 − 7 n1 − 7 n1 − 17 10
= − +1= +1
n1 − 17 n1 − 17 n1 − 17 n1 − 17
n1 − 14 n1 − 14 n1 − 24 10
= − +1= +1
n1 − 24 n1 − 24 n1 − 24 n1 − 24
Wow, so we get n1 − 17, n1 − 24 | 10. So we have two divisors of 10 that differ by n1 − 17 − (n1 − 24) = 24 − 17 = 7.
We can list them out - there are only 8:
−10, −5, −2, −1, 1, 2, 5, 10
and easily find the only such ones are {2, −5}, {−2, 5}. Thus, we get that either n1 − 17 = 2 or n1 − 17 = 5, so
n1 = 19, 22. By what we mentioned above, it’s easy to see that n2 = 19, 22, so
n1 · n2 = 19 · 22 = 418 .
So this problem showed us that polynomial problems also can be solved with some Number Theory. For fun,
however, solve the following parody of the AIME problem. Don’t solve it with number theory - use Zero
Polynomial Corollary to explicitly find P (x) and then solve it:
Exercise 6.4 (Parody of 2005 AIME II Problem 13). Let P (x) be a monica quadraticb polynomial with
integer coefficients that satisfies P (17) = 10 and P (24) = 17. Given that P (n) = n + 3 has two distinct
integer solutions n1 and n2 , find the product n1 · n2 . Hints: 53
The problem is that there aren’t many problems on this topic. Here’s a problem I made:
Exercise 6.5. Consider a polynomial f (x) with integer coefficients such that for any integer n > 0, f (n)−f (0)
is a multiple of the sum of the first n positive integers. Find the minimum value of f (2020) − f (0), given
that f (n + 1) > f (n) for all positive integers n. Hints: 45
a
This means the leading coefficient is 1.
b
This means the degree is 2.
§7 Advanced Algebraic Manipulations

This section was only created due to suggestions from freeman66 and ab xy123 on the AoPS Community.
Sometimes, when we are working with σk (see Vieta’s Formulas) for small numbers, we will need to do a
24
bunch of manipulations. It isn’t very easy to go back to Newton’s Formulas every time, so here is a list for
3 variables (we get 2 variables is too easy, four variables is uncommon), due to freeman66. This list is not
exhaustive, but it includes almost everything seen in an AIME setting. Also note that he uses the shorthand
u = σ1 , v = σ2 , w = σ3 .
1. a2 + b2 + c2 = u2 − 2v
Solution.
2. a3 + b3 + c3 = u(u2 − 3v) + 3w
3. a2 b2 + b2 c2 + c2 a2 = v 2 − 2uw
4. a4 + b4 + c4 = (u2 − 2v)2 − 2(v 2 − 2uw) = u4 − 4u2 v + 2v 2 + 4uw
5. (a + b)(b + c)(c + a) = uv − w
P
6. ab(a + b) = uv − 3w
cyc
7. (1 + a)(1 + b)(1 + c) = 1 + u + v + w
P
8. (1 + a)(1 + b) = 3 + 2u + v
cyc
(1 + a2 ) = u2 + v 2 + w2 − 2uw − 2v + 1
P
9.
cyc
I do suggest you go through these again to check that these are actually true. But let’s see some uses of these
manipulations. Most of them are AIME-like (suggested to me by ab xy123) and require clever insights of different
theorems to solve the problem:
Example 7.1 (PRMO 2019/2)

Let f (x) = x2 + ax + b. If for all nonzero real x

1 1
f x+ = f (x) + f
x x
and the roots of f (x) = 0 are integers, what is the value of a2 + b2 ?
This problem looks pretty hard - how are we going to solve it? Well, what is something that makes
1 1
x, , x +
x x
all look nice? I know that the only integer x such that x1 is also an integer is ±1. We can try plugging in x = 1.
We get
f (1 + 1) = f (1) + f (1)
We can use the polynomial to get
4 + 2a + b = f (2) = 2f (1) = 2(1 + a + b) = 2 + 2a + 2b
which gives b = 2. That’s a good start, but how do we find a? Well, let’s look at the problem. It says all roots
are integers. So what does this remind me of? Personally, I go back to the Rational Root Theorem. So we get
the roots are ±1, ±2. By Vieta’s Formulas, we also get the roots multiply to 2. So thus we get they are (1, 2) or
(−1, −2).
25
We got two solutions! How is this possible! Well, it just boils down to how attentive you are. We need to find a2
(because we already have b2 ), and a = 3, −3 are our two values of a (we used Vieta’s Formulas). So then, it’s
obvious - the sum is a2 + b2 = 13 no matter what! So the problem was crafted in such a way it’s impossible to
find the roots - you must find a2 .
Let’s see another example in action - we will use some of freeman66’s tactics:
Example 7.2 (Modified from RMO 2013/2)

Let f (x) = x3 + ax2 + bx + c and g(x) = x3 + bx2 + cx + a, where a, b, c are real numbers with c 6= 0.
Suppose that the following conditions hold:
• f (1) = 0
• the roots of g(x) = 0 are the squares of the roots of f (x) = 0.
Find the value of a2013 + b2013 + c2013 .
Solution. Well, let’s see what we know. We let the roots of f be r, s, t. By Vieta’s Formulas, we get
• r + s + t = −a
• rs + rt + st = b
• rst = −c
Now, the roots of g(x) are r2 , s2 , t2 , we get
• r2 + s2 + t2 = −b
• r2 s2 + r2 t2 + s2 t2 = c
• r 2 s2 t2 = a
Now, we can use some of freeman66’s properties. Specifically, using property 1, we get
a2 − 2b = r2 + s2 + t2 = −b
so b = a2 . Using property 3, we get
b2 − 2ac = b2 − 2(−a)(−c) = r2 s2 + r2 t2 + s2 t2 = c
so
a4 = b2 = (2a + 1)c
This means that
a4
c= (Θ)
2a + 1
Finally, we get that
c2 = r2 s2 t2 = −a
so we get (a, b, c) = (−c2 , c4 , c). But what about (Θ)? Can we use it? Trying to use it, we get
c8
c= =⇒ c7 = 1 − 2c2 (Π)
1 − 2c2
26
Oh god! Not a degree 7 equation. But we can still use it to check solutions, in case. Let’s see the other
information that we got in the problem. We know f (1) = 0. We know
f (x) = x3 − c2 x2 + c4 x = c
so
0 = f (1) = 1 − c2 + c + c4 = (c + 1)(c3 − c2 + 1)
But the second factor isn’t very nice, right? Let’s see what we get from (Π). We can factor it to get
0 = c7 + 2c2 − 1 = (c + 1)(c6 − c5 + c4 − c3 + c2 + c + 1)
so obviously c = −1 is a root. If we assume that r is a root of the second factor, then r3 = r2 − 1. We can
probably manipulate this, right? I mean, we must have that
r6 − r5 + r4 − r3 + r2 + r + 1 = 0
Let’s use our identity we found repeatedly. We get r6 = r5 − r3 , so plugging it in gives
r4 − 2r3 + r2 + r + 1 = 0
Again! We get r4 = r3 − r, so thus plugging it in gives
−r3 + r2 + 1 = 0
One last time! r3 = r2 − 1, so

2r2 = −2
which means r = ±i. But it’s easy to check that it isn’t a root of c3 − c2 + 1 = 0! So we must have c = −1.
Then, a = −1 and b = 1. Then, the problem becomes −1 − 1 + 1 = −1 .
Remark 7.3. The problem only asked the question if a, b, c are integers. We proved they must be!
Here are some other exercises:
Exercise 7.4 (1991 INMO Problem 2). How many ordered triples (x, y, z) of real numbers satisfy the
system of equations
x2 + y 2 + z 2 = 9,
x4 + y 4 + z 4 = 33,
xyz = −4?
Hints: 76
Exercise 7.5 (AoPS Forums).

x+y+z =1
x2 + y 2 + z 2 = 2
x3 + y 3 + z 3 = 3
Evaluate x4 + y 4 + z 4 . Hints: 18
27
§8 Worked Through Problems

Remember the problems I promised you at the beginning? Let’s look at them here:

Let P (x) be a nonzero polynomial such that (x−1)P (x+1) = (x+2)P (x) for every real x, and (P (2))2 = P (3).
Then P ( 27 ) = m
n , where m and n are relatively prime positive integers. Find m + n.
7
Solution. Eh, this problem looks too mild. Let’s spice it up! How about we find P (x), and I leave plugging in 2
for you?
First, let’s try to scout for roots. We note that x = −1 and x = −2 give ”easy” roots; here we get P (1) =
P (−1) = 0 (do the work here, I may be lying). Now, plugging in x = 0 gives us 0 = −P (1) = 2P (0), so 0 is also
a root.
So, what other roots are there, if any? Let’s consider a root r 6= 0, ±1. Then, what can we get? We plug them
into our only thing we know to get
(r − 1)P (r + 1) = (r + 2)P (r) = 0
but as r 6= 1, we have r + 1 = 0. Similarly, we get that r + 2, is a root, and so on.
Or really? What if r = −2? Then we sort of hit a stopping point - we can’t go forwards anymore, as we stop at
1. However, we only went forwards - we can do the same thing backwards to get
0 = (r − 2)P (r) = (r + 1)P (r − 1)
so as r 6= −1, we are good as we go backwards. Obviously, no nonzero polynomial has ∞ roots, so the only
roots are 0, ±1. Thus, we almost have our polynomial - P (x) = cx(x − 1)(x + 1). How to find c? Well, there’s a
reason we have that second equation. That gives
P (2) = c · 2 · 1 · 3 = 6c
P (3) = c · 3 · 2 · 4 = 24c
so thus
36c2 = P (2)2 = P (3) = 24c
As c is nonzero, we divide by 0 to get c = 23 . Thus, P (x) = 23 x(x − 1)(x + 1). I’ll do a formal write-up in the
following:
Formal Proof.
Call the assertion Q(x) as (x − 1)P (x + 1) = (x + 2)P (x)a . Then, we get that 0, ±1 are roots from
Q(1), Q(−2), and Q(0).
Now, we shall prove that these are the only such roots. Assume that r is another root. If r is not a negative
integer, we note that Q(r) gives
(r − 1)P (r + 1) = (r + 2)P (r) = 0
28
so as r − 1 6= 0, we have that r + 1 is also a root. Now, we shall induct on k to show that r + k is a root:
Base Case. k = 1 has been shown above.

Induction Hypothesis. Assume it is true for some k; we will show it for k + 1.
Induction Step. We have that if r is not an integer, r + k + 1 can not be an integer, so it can not be 0, ±1.
If integer r ≥ 2, then r + k + 1 ≥ 2 and thus r 6= 0, ±1. Thus, we have that in this case r + k + 1 is an
integer by applying our Base Case to r + k instead of r.
For negative integers, by using Q(r − 1), we get
0 = (r − 2)P (r) = (r + 1)P (r − 1)
so for negative integer roots r not −1, we get r − 1 is also a root. Now, we shall induct on k to show that
r − k is a root:
Base Case. k = 1 has been shown above.
Induction Hypothesis. Assume it is true for some k; we will show it for k + 1.
Base Case. As r ≤ −2, then r + k + 1 ≤ −2 and thus r 6= 0, ±1. Thus, we have that in this case r − (k + 1)
is an integer by applying our Base Case to r − k instead of r.
Now, in either case we get infinitely many roots, so by Zero Polynomial Corollary, we have that P (x) is
the zero polynomial, a contradiction. Thus, we get that P (x) = cx(x − 1)(x + 1). Plugging in x = 2, 3, we
have that P (2) = 6c, P (3) = 24c, so
36c2 = P (2)2 = P (3) = 24c
and as c is nonzero, c = 32 . Thus, we get that P (x) = 23 x(x − 1)(x + 1) and in particular

m 7 2 7 9 5 105
=P = · · · =
n 2 3 2 2 2 4
which implies m + n = 109 .

a
That’s so I don’t have to keep writing awkwardly our initial constraint

Determine w2 + x2 + y 2 + z 2 if
x2 y2 z2 w2
+ + + =1
22 − 1 22 − 32 22 − 52 22 − 72
x2 y2 z2 w2
+ 2 + 2 + 2 =1
42 − 1 4 − 32 4 − 52 4 − 72
x2 y2 z2 w2
+ 2 + 2 + 2 =1
62 − 1 6 − 32 6 − 52 6 − 72
x2 y2 z2 w2
+ 2 + 2 + 2 =1
82 − 1 8 − 32 8 − 52 8 − 72
Solution. Ok, now this looks daunting. And in the wrong handout. Where’s the polynomial? But that’s the
beauty of this problem - the polynomial comes when we see that trying to expand is too tedious. So, first, the
29
w2 , x2 , y 2 , z 2 seems 150% unecessary - like we could replace them with a, b, c, d. So let’s do that:
a b c d
+ + + =1
22 −1 22
− 32 22
− 52 22
− 72
a b c d
+ 2 + 2 + 2 =1
42 − 1 4 − 32 4 − 52 4 − 72
a b c d
+ 2 + 2 + 2 =1
62 − 1 6 − 32 6 − 52 6 − 72
a b c d
2
+ 2 + 2 + 2 =1
8 −1 8 − 32 8 − 52 8 − 72
It’s already nicer. We just have to find a + b + c + d. Ok, now what else happens nicely? The denominators are
always t2 − 1, t2 − 9, t2 − 25, t2 − 49 for t = 2, 4, 6, 8. Hmmm...if we used this, we would get really big numbers,
like degree 8 polynomials. What about t − 1, t − 9, t − 25, t − 49, where t = 4, 16, 36, 64.
Remark 8.3. It’s more beneficial to spend time to step back and look at a problem from a different angle, especially
when you have time and you realize the numbers you are getting are cancerous. Also, try to show any pattern you see
holds.
Now, let’s make that substitution, so we get

a b c d
+ + + =1
t − 1 t − 9 t − 25 t − 49
for t = 4, 16, 36, 64. So what can we do? Well, we can’t deal with these, but what if we made it a polynomial?
Then, we would need a common denominator. Thus, we get
a(t − 9)(t − 25)(t − 49) + b(t − 1)(t − 25)(t − 49) + c(t − 1)(t − 9)(t − 49) + d(t − 1)(t − 9)(t − 25)
=1
(t − 1)(t − 9)(t − 25)(t − 49)
For the sake of allowing this to fit in the page, define
τa = (t − 9)(t − 25)(t − 49) = t3 − 83t2 + 1891t − 11025
τb = (t − 1)(t − 25)(t − 49) = t3 − 75t2 + 1299t − 1225

τc = (t − 1)(t − 9)(t − 49) = t3 − 59t2 + 499t − 441
τd = (t − 1)(t − 9)(t − 25) = t3 − 35t2 + 259t − 225
Remark 8.4. It is implied that these are polynomials. I should do a better job of this, but it’s in the formal write-up.
Then, cross multiplying would give us
aτa + bτb + cτc + dτd = (t − 1)(t − 9)(t − 25)(t − 49)
for t = 4, 16, 36, 64. How can we manipulate this? Well, the equation has degree 4, and we have four roots, so we
can use the Unique Factorization of Polynomials to say they are equal. But which one of aτa + bτb + cτc + dτd
and (t − 1)(t − 9)(t − 25)(t − 49) do we choose?
What are we doing? We have to write f (t) = 0, so we want the polynomial of their difference,
aτa + bτb + cτc + dτd − (t − 1)(t − 9)(t − 25)(t − 49) = 0 (#)
30
and get that t = 4, 16, 36, 64 are roots. So then, we can take this equal to (t − 4)(t − 16)(t − 36)(t − 64), right?
But what about the leading coefficients? Don’t forget them! We see that τa , τb , τc , τd are all degree 3 polynomials,
so (#) has leading coefficient −1. Thus, we actually get that (#) = −(t − 4)(t − 16)(t − 36)(t − 64), so
aτa + bτb + cτc + dτd − (t − 1)(t − 9)(t − 25)(t − 49) = −(t − 4)(t − 16)(t − 36)(t − 64)
Yeah, so we can look at what we want to find: a + b + c + d. As this has infinitely many roots, we have that by
Zero Polynomial Corollary, the coefficients on each side are equal. But, what can we do?
Looking at τa , τb , τc , τd , we get that as we move towards lower degree terms, the polynomial gets more “cancerous”,
shall we say? So thus, we start with the highest degree terms: 4. We get that the left hand side has leading
coefficient −1 (τa , τb , τc , τd all have degree 3), but the right hand side has coefficient −1.
Oh well, let’s go to degree 3. τa has the coefficient of degree 3 as 1, so aτa has coefficient a. Similarly, bτb , cτc , dτd
all have coefficients of t3 as b, c, d. Now, by Vieta’s Formulas, the coefficient of t3 in (t − 1)(t − 9)(t − 25)(t − 49)
is −(1 + 9 + 25 + 49) = −84. However, once again by Vieta’s Formulas, the coefficient of t3 in −(t − 4)(t −
16)(t − 36)(t − 64) is 4 + 16 + 36 + 64 = 120 (note the negative signs cancelled out). Thus, we get that
a + b + c + d − (−84) = 120 =⇒ a + b + c + d = 120 − 84 = 36 .
Once again, I’ll provide a formal write-up:
Formal Proof.
Define a = x2 , b = y 2 , c = z 2 , d = w2 , then we get that the problem condition rewrites to
a b c d
+ + + =1
22 −1 22
− 32 22
− 52 22
− 72
a b c d
+ 2 + 2 + 2 =1
42 − 1 4 − 32 4 − 52 4 − 72
a b c d
+ 2 + 2 + 2 =1
62 − 1 6 − 32 6 − 52 6 − 72
a b c d
2
+ 2 + 2 + 2 =1
8 −1 8 − 32 8 − 52 8 − 72
and we need to find w2 + x2 + y 2 + z 2 = a + b + c + d. Notice that

a b c d
+ + + =1 (!)
t − 1 t − 9 t − 25 t − 49
holds for t = 4, 16, 36, and 64 as guaranteed by the problem statement. Thus, we define
τa (t) = (t − 9)(t − 25)(t − 49)
τb (t) = (t − 1)(t − 25)(t − 49)

τc (t) = (t − 1)(t − 9)(t − 49)
τd (t) = (t − 1)(t − 9)(t − 25)
τ (t) = (t − 1)(t − 9)(t − 25)(t − 49)
31
Now, we get that

a a(t − 9)(t − 25)(t − 49) aτa (t)
= =
t−1 (t − 1)(t − 9)(t − 25)(t − 49) τ (t)
Thus, by symmetry, we get that
b bτb (t)
=
t−9 τ (t)
c cτc (t)
=
t − 25 τ (t)
d dτd (t)
=
t − 49 τ (t)
Thus, (!) asserts that
aτa (t) + bτb (t) + cτc (t) + dτd (t)

= 1 =⇒ aτa (t) + bτb (t) + cτc (t) + dτd (t) = τ (t)
τ (t)
where t = 4, 16, 36, 64. Thus, we get that
χ(t) = aτa (t) + bτb (t) + cτc (t) + dτd (t) − τ (t) (Ω)
has zeroes at t = 4, 16, 36, 64. Furthermore, as τa , τb , τc , τd have degree 3 and τ has degree 4, we have that
χ has degree 4. Thus, we can write (by Unique Factorization of Polynomials):
χ(t) = k(t − 4)(t − 16)(t − 36)(t − 64) (ζ)
for some (nonzeroa ) real k. Now, we get that comparing the leading coefficients, by (Ω), the leading
coefficient of χ is the opposite of the leading coefficient of τ (as τa , τb , τc , τd have degrees all less than τ ).
Thus, because the leading coefficient of τ is 1, the leading coefficient of χ is −1, and thus so is k (by (ζ)).
Now, we compare the coefficients of t3 . The coefficient of t3 in τa , τb , τc , τd is 1. The coefficient of t3 in τ is
−(1 + 9 + 25 + 49) = −84
so thus the coefficient of t3 in χ is

a + b + c + d + 84
However, the coefficient of χ in −(t − 4)(t − 16)(t − 36)(t − 64) is
−(4 + 16 + 36 + 64) = 120
These are equal, so

a + b + c + d + 84 = 120 =⇒ a + b + c + d = 36
a
This is as τ is the only polynomial with degree 4, so the x4 term can not “cancel” out.
§9 Parting Words and Final Problems

So with this, you should be able to solve almost any AIME Problem on polynomials. I hope this document
helped you learn a bit about how to use polynomials in all kinds of contexts, even ones that aren’t obviously
apparent. Any suggestion would be extremely helpful, whether it would be problem suggestions, mistakes I
made, or stuff I should explain better. Here’s a final problem set that should incorporate (almost) every AIME
Problem which requires polynomials (that hasn’t been solved above). In addition, there are other problems,
which are suggestions from one of twinbrian or ab xy123:
32
Problem
√ 9.1 (1983 AIME Problem 3). What is the product of the real roots of the equation x2 + 18x + 30 =
2 x2 + 18x + 45? Hints: 63
Problem 9.2 (2013

√
AIME I Problem 5). The real root of the equation 8x3 − 3x2 − 3x − 1 = 0 can be written
√
3 3
a+ b+1
in the form c , where a, b, and c are positive integers. Find a + b + c. Hints: 64
Problem 9.3 (2010 AIME I Problem 6). Let P (x) be a quadratic polynomial with real coefficients satisfying
x2 − 2x + 2 ≤ P (x) ≤ 2x2 − 4x + 3 for all real numbers x, and suppose P (11) = 181. Find P (16). Hints: 66
Problem 9.4 (2015 AIME II Problem 6). Steve says to Jon, “I am thinking of a polynomial whose roots are
all positive integers. The polynomial has the form P (x) = 2x3 − 2ax2 + (a2 − 81)x − c for some positive integers
a and c. Can you tell me the values of a and c?”
After some calculations, Jon says, “There is more than one such polynomial.”
Steve says, “You’re right. Here is the value of a.” He writes down a positive integer and asks, “Can you tell me
the value of c?”
Jon says, “There are still two possible values of c.”
Find the sum of the two possible values of c. Hints: 49

1 1
Problem 9.5 (2016 AIME II Problem 6). For polynomial P (x) = 1 − x + x2 , define
3 6
50
X
Q(x) = P (x)P (x3 )P (x5 )P (x7 )P (x9 ) = ai xi .
i=0
50
P m
Then |ai | = , where m and n are relatively prime positive integers. Find m + n. Hints: 8
i=0 n
Problem 9.6 (2004 AIME I Problem 7). Let C be the coefficient of x2 in the expansion of the product
P (x) = (1 − x)(1 + 2x)(1 − 3x) · · · (1 + 14x)(1 − 15x). Find |C|. Hints: 5 55
Problem 9.7 (2011 AIME I Problem 7). Find the number of positive integers m for which there exist nonnegative
integers x0 , x1 , . . . , x2011 such that
2011
X
m x0 = mxk .
k=1
Hints: 3
Problem 9.8 (1984 AIME Problem 8). The equation z 6 + z 3 + 1 = 0 has complex roots with argument θ
between 90◦ and 180◦ in the complex plane. Determine the degree measure of θ. Hints: 12
Problem 9.9 (1989 AIME Problem 8). Assume that x1 , x2 , . . . , x7 are real numbers such that
x1 + 4x2 + 9x3 + 16x4 + 25x5 + 36x6 + 49x7 = 1

4x1 + 9x2 + 16x3 + 25x4 + 36x5 + 49x6 + 64x7 = 12
9x1 + 16x2 + 25x3 + 36x4 + 49x5 + 64x6 + 81x7 = 123.
Find the value of 16x1 + 25x2 + 36x3 + 49x4 + 64x5 + 81x6 + 100x7 . Hints: 52
√
Problem 9.10 (2014 AIME I Problem 9). Let x1 < x2 < x3 be the three real roots of the equation 2014x3 −
4029x2 + 2 = 0. Find x2 (x1 + x3 ). Hints: 15
33
Problem 9.11 (2010 AIME II Problem 10). Find the number of second-degree polynomials f (x) with integer
coefficients and integer zeros for which f (0) = 2010. Hints: 6
Problem 9.12 (2015 AIME I Problem 10). Let f (x) be a third-degree polynomial with real coefficients satisfying
|f (1)| = |f (2)| = |f (3)| = |f (5)| = |f (6)| = |f (7)| = 12.
Find |f (0)|. Hints: 73
Problem 9.13 (2019 AIME I Problem 10). For distinct complex numbers z1 , z2 , . . . , z673 , the polynomial
(x − z1 )3 (x − z2 )3 · · · (x − z673 )3
can be expressed as x2019 + 20x2018 + 19x2017 + g(x), where g(x) is a polynomial with complex coefficients and
with degree at most 2016. The value of

X

z j z k

1≤j<k≤673
m
can be expressed in the form n, where m and n are relatively prime positive integers. Find m + n. Hints: 32
Problem 9.14 (1986 AIME Problem 11). The polynomial 1 − x + x2 − x3 + · · · + x16 − x17 may be written in
the form a0 + a1 y + a2 y 2 + · · · + a16 y 16 + a17 y 17 , where y = x + 1 and the ai ’s are constants. Find the value of
a2 . Hints: 4
Problem 9.15 (1988 AIME Problem 13). Find a if a and b are integers such that x2 − x − 1 is a factor of
ax17 + bx16 + 1. Hints: 2
Problem 9.16 (2000 AIME √ Problem 13). The equation 2000x6 + 100x5 + 10x3 + x − 2 = 0 has exactly two
real roots, one of which is m+r n , where m, n and r are integers, m and r are relatively prime, and r > 0. Find
m + n + r. Hints: 74 30
Problem 9.17 (2004 AIME I Problem 13). The polynomial P (x) = (1 + x + x2 + · · · + x17 )2 − x17 has 34 complex
roots of the form zk = rk [cos(2πak ) + i sin(2πak )], k = 1, 2, 3, . . . , 34, with 0 < a1 ≤ a2 ≤ a3 ≤ · · · ≤ a34 < 1 and
rk > 0. Given that a1 + a2 + a3 + a4 + a5 = m n , where m and n are relatively prime positive integers, find m + n.
Hints: 68
Problem 9.18 (2007 AIME II Problem 14). Let f (x) be a polynomial with real coefficients such that f (0) = 1,
f (2) + f (3) = 125, and for all x, f (x)f (2x2 ) = f (2x3 + x). Find f (5). Hints: 11 38 9
Problem 9.19 (2020 AIME I Problem 14). Let P (x) be a quadratic polynomial with complex coefficients whose
x2 coefficient is 1. Suppose the equation P (P (x)) = 0 has four distinct solutions, x = 3, 4, a, b. Find the sum of
all possible values of (a + b)2 . Hints: 37 58 19
Problem 9.20 (2011 AIME I Problem 15). For some integer m, the polynomial x3 − 2011x + m has the three
integer roots a, b, and c. Find |a| + |b| + |c|. Hints: 36 77
Problem 9.21 (1984 USAMO Problem 1). In the polynomial x4 − 18x3 + kx2 + 200x − 1984 = 0, the product
of 2 of its roots is −32. Find k. Hints: 70
x
Problem 9.22 (2017 RMO Problem 3). Let P (x) = x2 + + b and Q(x) = x2 + cx + d be two polynomials with
2
real coefficients such that P (x)Q(x) = Q(P (x)) for all real x. Find all real roots of P (Q(x)) = 0. Hints: 56 23
Problem 9.23 (2018 PRMO Problem 30). Let P (x) = a0 + a1 x + a2 x2 + · · · + an xn be a polynomial in which
ai is non-negative integer for each i ∈ 0, 1, 2, 3, ...., n . If P (1) = 4 and P (5) = 136, what is the value of P (3)?
Hints: 7 75 59
34
Problem 9.24 (2020 February HMMT Algebra and Number Theory Problem 8). Let P (x) be the unique
polynomial of degree at most 2020 satisfying P (k 2 ) = k for k = 0, 1, 2, . . . , 2020. Compute P (20212 ). Hints: 42
41 46
Problem 9.25 (Modified from 2016 PUMaC A7). Let SP be the set of all polynomials P with complex
coefficients, such that P (x2 ) = P (x)P (x − 1) for all complex numbers x. Hints: 29 65 21 25
35
§10 Hints
1. Find σ1 , σ2 , σ3 . This should be a routine exercise, and then consider all roots of x3 − σ1 x2 + σ2 x − σ3 = 0.
2. You could try to use recursions if you recognize x2 − x − 1. Otherwise, try to use polynomial division with smaller
sequences - and see if there is a pattern.
3. Doesn’t seem like a polynomial problem? Think of a simple value of which you know xk for all k - and find P (x).
Then, try to use polynomial division. The backwards of this, however, is purely number theory. I’m sure you can do
it!
4. Try a geometric series as the value of the polynomial. It should make the rest very easy.
1 1
5. Well, considering the x2 term isn’t very good. How about the polynomial with roots r1 , r2 , . . . (r1 , r2 , . . . are the
roots of P (x))? Is it simpler now?
6. Write it in factored form! Where can you go from there?
7. Can you bound n? Remember, 53 = 125!
8. This is more daunting than it looks - try to see what value of x satisfies P (x) is the sum of the coefficients of P (x).
Can you generalize to Q(x)?
9. Consider the polynomial g(x) with the roots found above, and write f (x) = g(x)h(x). Does h(x) satisfy the same
constraints as f (x)? Use this to finish off the problem (after finding the form of f (x)).
10. Use Newton’s Sums, but make sure to factor first! Make it as simple as possible! (In general, try to keep everything
factored for as long as possible).
11. Try to find a root with r > 1. Can you find another? What is the magnitude of that? Greater, equal to, or less than
r? Can you keep going? Remember Zero Polynomial Corollary and use ∞ roots.
12. Let x = z 3 . Does this look familiar (third roots of unity)? Use the polar form to get such a value of θ.
13. Use induction and the fact nk + k+1 n
= n+1
9
k .
14. Use Complex Conjugate Theorem as well as the fact that every polynomial of odd degree has at least one real root.
We can show the last statement as follows: graph f (x), and note on one side, it goes to ∞ while on the other it goes
to −∞ so it must hit the x−axis in the middle.
√
15. Let n = 2014. Write 2014 and 4029 in terms of n. Now try to factor!
16. How many roots are there? Can you use a double counting method?
17. Can you use Vieta’s to find an expression for t in terms of r and s? Then try to expand!
18. Try to use manipulations 2 and 4 from freeman66’s list. It’s not 4!
19. For some of the cases, you may have to solve for a and b explicitly. There’s nothing to do about that.
20. What’s the imaginary part of r?
21. What if |r| < 1? First, however, ignore r = 0.
22. Use the exponent laws - don’t try to convert into complex numbers.
23. Expanding is good. We already should know one value, so there are only two “unknowns”.
√
1+ 3i
24. Rewrite 2 as a root of unity. Can you simplify, for example, z 2018 to a lower power?
25. Assume |r| = 1, and let r = a + bi. You’ll get an equation in terms of a and b. Expand and solve!
26. The answer to “do the results seem familiar” should be yes. They should relate to the coefficients in the expansions.
27. Try to expand f · g and f + g!

28. Assume z = a + bi, w = c + di, and then expand both sides and then show they are equal.
9 n

We use k
to mean the number of ways to choose an (unordered) set k elements from n elements.
36
29. What are the roots of P (x2 ) in terms of the roots of P (x)?
30. Only one of the two factors will have real roots. Use the Discrimant of a Quadratic Polynomial to help you determine
this (if x2 is not real, neither is x).
31. What’s the third root? Also use Factor Theorem; it can be helpful to reduce the amount of extra algebra done. But
alas, the problem ends bashy.
32. It’s a little more complicated than straightforward Vieta’s Formulas - can you also use Newton’s Formulas (or
freeman66’s tactics)?
33. Assume z = a + bi, and then expand both sides and then show they are equal.
35. Use the Rational Root Theorem to find all possible roots of both polynomials.
36. Without loss of generality, let |a| ≤ |b| ≤ |c|. Consider what c is in terms of a and b. Also use Newton’s Formulas to
reduce your search.
37. Consider the roots of P (x) as r1 , r2 . How many cases do you really have here for P (x) = r1 , P (x) = r2 . Remember
to exploit symmetry.
38. Repeat the same thing with r < 1. Can you find a root with |r| = 1?
39. Use the first equation to substitute.
40. Expand! No better hint for this problem.
41. Now, here’s the magic (it’s called finite differences). Consider Q(x + 1) − Q(x).
• What’s its degree?
• List 4038 roots.
• Can you find the value of the leading coefficient (what is Q(1) − Q(−1))?
42. If you know Lagrange Interpolation, use it! Otherwise, can you find an obvious root of P (x)? Can you consider a
polynomial Q(x) (of degree 4039) with roots k = 1, 2, . . . , 2020 and manageable values for k = −1, −2, . . . , −2020?
43. Solve the exercise in the order given ρ1 , ρ2 , ρ3 . You may need to substitute.
44. Try using y = 2111x . Then, try to also work backwards as we did in the above problem to find a nice expression for
the sum of all values for x in terms of all values of y.
45. Remember, 2020 also divides f (2020) − f (0). Now, can you find a polynomial that satisfies f (2020) as the value you
had. I suggest looking at the formula for the sum of the first n numbers.
46. If you don’t recognize some of the scary terms - think of binomial coefficients.
47. Can you experiment with some smaller roots? Maybe use Rational Root Theorem?
48. Look at Complex Conjugate Theorem for a weakened version. Follow the same proof.
49. Take that problem step by step. Also, Vieta’s Formulas comes in handy here. And Newton’s Formulas can help too.
Alternatively, consider #1 from freeman66’s tactics.
50. This has a nice polynomial solution. Note a number is real if and only if it is equal to its conjugate. Use that to
obtain a polynomial in z. How many roots does it have?
51. Use Complex Conjugate Theorem and the fact that every polynomial of odd degree has at least one real root.
52. Try to generalize (like we did in AIME 1984/15)! Can you find a polynomial here? Can you find the coefficients?
53. Try to find a linear polynomial Q(x) with Q(17) = 10 and Q(24) = 17. Then factor P (x) − Q(x).
54. U
55. Newton’s Formulas can help too. Alternatively, consider #1 from freeman66’s tactics.
37
56. Assume P (x) has real roots (otherwise the problem is trivial). Consider any root r of P (x). What’s P (r)Q(r)?
Q(P (r))?
57. How many roots are in common between Q1 (x) and Q2 (x)?
58. Also remember that the polynomials P (x) − r1 , P (x) − r2 have the same linear term. Some cases will be nearly
trivial with Vieta’s Formulas. Others will be almost impossible without bashing.
59. The bounds should be easy to obtain for the rest except a0 , a1 , where you will get two equations in two variables.
Elimination works, and so does substitution, and any other method taught in Algebra I. Solve for P (x).
60. How can you relate the roots of Pk and Pk+1 . The rest should follow from Vieta’s relationships.
61. It was k1 for σ1 , k1 k2 for σ2 , and k1 k2 k3 for σ3 . See a pattern?
63. Make the substitution y = x2 + 18x + 30 and solve for y instead.
64. Do some of the terms look similar to those in (x + 1)3 ? What can you conclude (rearrange it to (x + 1)3 = something)?
65. Consider a root r. What can you tell me if |r| > 1 (specifically, √
what is the magnitude in relationship to r)?
Remember Zero Polynomial Corollary with ∞ roots. Also use m > m if m > 1.
66. Suppose Q(x) = x2 − 2x + 2 and R(x) = 2x2 − 4x + 3. When Q(x) = R(x), do we know the value of P (x)? Can you
find the exact value of P (x) from there?
P
67. Like in the above two problems, just break the into less daunting sums and take casework.
68. Use a geometric series, and then expand! It won’t disappoint if you try to factor.
69. Can you try to factor? Use the fifth roots of unity to help you.
70. The algebraic manipulation ab + ac + ad + bc + bd + cd = (a + b)(c + d) + ab + cd comes in handy a lot.
71. Factor it! How many roots does it have?
72. Use (β) and (Γ).
73. This was built for Vieta’s Formulas. Find the roots of P (x) + 12 and P (x) − 12 (note the only change between the
polynomials is the constant term).
74. You could try to factor this. Focus on the terms of even degree and odd degree separately.
75. Is it even possible that am = 0, where m is the maximum value of n? Remember that the sum of the ai is 4! Can it
be more than 1?
76. Try to use manipulations 2 and 4 from freeman66’s list.
77. Take cases on if c > 0 or c < 0. Bound c well, and do casework on the rest.
78. Use Euler’s Formula.
79. Try to rewrite 31 = 52 + 5 + 1. Can you use roots of unity (you may have to factor xn − 1)?
80. Let ζ = sin2 x and χ = cos2 x. Do you have any information on ζ and χ that would allow you to solve for ζχ?
38
§A Appendix A: Proof of Results

Let’s prove some of the results here:
Theorem A.1 (Fundamental Theorem of Algebra)

coefficients), there exists a root r ∈ C (aka f (r) = 0).
Proof due to Matthew Steed, University of Chicago. We shall use Liouville’s Theorem, which is a powerful argu-
ment in complex analysis which states the following:
Theorem A.2 (Liouville’s Theorem)

Every bounded holomorphic function must be constant.
That’s it. It’s pretty powerful, and for sake of completeness, I include a proof:
Proof. Note any holomorphic function f is analytic. Consider the Taylor Series about 0:
∞
X
f (x) = bk xk
k=0
Then, using Cauchy’s Integral Formula, we get
f (k) (0) 1 f (ζ)

I
bk = = dζ
k! 2πi C ζ k+1
where C is a circle radius r (arbitrary) centered at 0. Now, suppose f is bounded. Then, if |f (x)| ≤ M , we get
1 |f (ζ)| M M 2πr M
I I
|bk | ≤ |dζ| ≤ |dζ| = = k
2π C |ζ|k+1 C r k+1 2πr k+1 r
Taking r → ∞, we accomplish our proof.
Back to the proof of Fundamental Theorem of Algebra. Consider a disk of radius R used in the previous proof.
There exists some α on the disk such that |f (α)| is aminimum on the disk. We suppose again that f (α) 6= 0.
1 1
For any z such that |z| ≥ |R|, |f (z)| > |f (α)|, so f (α) > f (z) . By Liouville’s Theorem, this is bounded above,

so f1 is constant, so |f | is constant, which is a contradiction. Thus, f (α) = 0.

Theorem A.3 (Fundamental Theorem of Symmetric Polynomials)

Any symmetric polynomial can be expressed as the sum/product of a bunch of different symmetric
polynomials.
Proof. We will do an induction on the degree of the polynomial m:
Base Case.
m = 1 is obvious.
39
Induction Hypothesis.
Assume the statement is true for all m − 1 ≥ k ≥ 1. We shall prove it is true for m.
Induction Step.
Now, for the rest of the proof to work, we can break up our polynomial into a bunch of symmetric polynomials
of the same degree. We shall thus focus only when the degree of each term is m. Now, let us order the
terms of our symmetric polynomial, with axa11 xa22 · · · coming before bxb11 xb22 · · · if and only if for the first k
such that bk 6= ak , ak > bk . Now, consider any term of our symmetric polynomial f = axa11 xa22 · · · xann (and
a1 + a2 + · · · + an = m. We can assume
a1 ≥ a2 ≥ · · · ≥ an
because every permuatation of a1 , a2 , . . . , an is included. Thus, we consider
g = aσ1a1 −a2 σ2a2 −a3 · · · σnan
Then, this has the same leading term as f , so applying our induction hypothesis, we can write f − g as
the sum/product of a bunch of symmetric polynomials, so we can write f as a sum/product of a bunch of
symmetric polynomials.
§B Appendix B: Polynomial Division

We introduce polynomial division and give a proof of the Remainder Theorem. Consider two polynomials, f (x)
and g(x). Then, we can write
f (x) = g(x)q(x) + r(x)
where deg r < deg g. We can prove the existence of such by induction on deg f :
Base Case.
When deg f = deg g, we consider q(x) the constant defined as the ratio of the leading coefficients of f to g.
Then, we get that f (x) − q(x)g(x) has it’s terms of deg g cancel out, so thus we get our claim is valid in
this case. Where deg f < deg g, we get that r(x) = f (x) and q(x) = 0 works.
We assume the result holds for all polynomials such that deg f ≤ k for some k. We shall show the result
holds for k + 1 as well.
Induction Step.
Consider p as the ratio of the leading coefficients of f to g. Then, we get that considering
f 0 (x) = f (x) − pxdeg f −deg g g(x)
we get the leading term (of degree deg f ) cancel, so thus deg f 0 ≤ k. Thus, we get
f 0 (x) = g(x)q(x) + r(x)
40
from our induction hypothesis, and then
f (x) = g(x)(q(x) + pxdeg f −deg g ) + r(x)
completing the induction step.
Now, we can also show uniqueness. We suppose that
f (x) = g(x)q1 (x) + r1 (x) = g(x)q2 (x) + r2 (x)
Then, we have that

g(x)(q1 (x) − q2 (x)) = r2 (x) − r1 (x)
Now, we note that if q1 6= q2 , we get that the right hand side has at least deg g roots by the Fundamental
Theorem of Algebra, but the left hand side has less than deg g roots by the Fundamental Theorem of Algebra (as
deg r1 , deg r2 < g by assumption). Thus, this is a contradiction, so thus q1 = q2 and r1 = r2 , so the expressibility
is unique.
But how do we do this? Remember normal division? We can use something like that. Let’s recap:
949
13 12345
117
64
52
125
117
8
We can do something similar with polynomials:

X 2 + 2X + 2
X 3 + X 2 + 0X − 1

X −1
− X3 + X2
2X 2 + 0X
− 2X 2 + 2X
2X − 1
− 2X + 2
1
Let’s go step by step. First, we take the smallest number of digits such that we can find a multiple of 13:
9
13 12345
117
6
Similarly, we do the same in polynomials long division:

X2
X 3 + X 2 + 0X − 1

X −1
− X3 + X2
2X 2 + 0X
41
Now, we just reiterate the process:
94
13 12345
117
64
52
12
X 2 + 2X + 2
X 3 + X 2 + 0X − 1

X −1
− X3 + X2
2X 2 + 0X
− 2X 2 + 2X
2X − 1
Until we get the quotient as described above. And for the sake of demonstration that it can be done with not
only linear polynomials:
X 2 + 2X + 1
X2 + X − 1 X 4 + 3X 3 + 2X 2 − 2X − 3

− X4 − X3 + X2
2X 3 + 3X 2 − 2X
− 2X 3 − 2X 2 + 2X
X 2 + 0X − 3
− X2 − X + 1
−X −2
But onto the proof of Remainder Theorem. We have that we can divide a polynomial f (x) by x − r to get
f (x) = (x − r)g(x) + h(x)
However, we get plugging in x = r, we get
f (r) = (r − r)g(r) + h(r) = h(r)
However, we note that deg h < deg(x − r) = 1, so deg h = 0, so h(x) is constant. Thus, h(x) = f (r), so the
remainder is indeed f (r).
§C Appendix C: Real Roots

This section is dedicated to how do we know where a real (possibly irrational) root is. The first theorem is called
the intermediate value theorem and is also used a lot in calculus:
Theorem C.1 (Intermediate Value Theorem)

Consider a continuous function f : I → R for some interval I = [a, b] (with a < b). Then, for all
c ∈ (f (a), f (b)), we can find some a < k < b such that f (k) = c.
42
Remark C.2. It isn’t necessary I = [a, b]. It can be [a, b), (a, b], (a, b) as well. That’s just how we talk about the
intermediate value theorem in normal context.
Proof. Consider the set S = {x | x ∈ I, f (x) ≤ c}, or all elements x in I such that f (x) ≤ c. Now, we note that
as min(f (a), f (b)) < c (by our assumption), so thus we can talk about the supremum of S, say k. We shall show
f (k) = c. Otherwise, by the definition of continuity, we can find for all > 0 some δ > 0 such that
|f (x) − f (k)| ≤ , |x − k| ≤ δ
Now, consider the interval (k − δ, k + δ). By our assumption, we have
f (x) − < f (k) < f (x) +
As f is the supremum, we must have for x ∈ (k − δ), we get
f (k) < f (x) + ≥ c +
and for x ∈ (k, k + δ), we get

f (k) > f (x) > > c −
so thus for arbitrary > 0, we have
c − < f (k) < c +
Now, by the Squeeze Theorem (or taking lim ), we get
→0
f (k) = c
which is the theorem statement.
Now, how does Intermediate Value Theorem help? Let’s take a look at 2005 AIME I Problem 8. In particular,
how did they know there were 3 real roots? Well, we note the function
f (x) = 2333x−2 + 2111x+2 − 2222x+2 − 1
is definitely continuous (essentially almost all functions are continuous, these can be visualized as being smooth
graphs). So we will try to find some values of f (x). We can make the following table:
f (−∞) -1
f (0) 1
f (1) Really Huge Number
so that won’t really work. We found guarantee of a root, but not 3. Let’s try the substitution that motivates
the solution of the problem:
1
g(y) = y 3 − 2y 2 + 4y − 1
4
1
If we can find three positive roots of this, we should be done, right (as for any solution y we can take x = 111 log2 x?
Let’s make a table (I’m only including even values so I get eve:
f (0) -1
f (2) 1
f (4) -1
f (6) 5
so we have a real root y in each of (0, 2), (2, 4), (4, 6), so we have three positive integer roots by Intermediate
Value Theorem. This is also very helpful when using the rational root theorem - sometimes it isn’t easy to do
the division - but it’s better to use this to find integer values and then find where there are roots for sure. Now,
let’s talk about this following magnificent result due Descartes:
43
Theorem C.3 (Descartes’ Rule of Signs)

Consider a polynomial (with deg f = n ≥ 1)
f (x) = an n xn + an−1 n−1 xn−1 + · · · + a0 0
where an > 0 and n ∈ {−1, 0, 1}. Let m be the number of times k k−1 = −1. Then, the number of positive
roots (say p) (counting multiplicities, i.e. the roots of (x − 1)2 are 1, 1) is at most m, and furthermore leaves
the same remainder as m when divided by 2.
Remark C.4. This following heuristic argument helped me understand the argument (due to Professor Stewart A.
Levin) - as x goes towards 0, the constant term comes into play, and as we move to +∞, the leading term comes into
play. In the middle the other ones have a chance to dominate, but sometimes are still dominated by other terms.
Proof due to Xiaoshen Wang. Divide by an n (as it doesn’t affect the product k k−1 as it is divided by 1) so
thus
f (x) = xn + an−1 n−1 xn−1 + · · · + a0 0
Now, if 0 = 0, then we can “remove” it as it doesn’t affect the number of positive roots. Thus, we also assume
0 6= 0. We have the following lemma:
Lemma C.5 — If 0 = 1, then p is even. Otherwise, p is odd.
Proof. We note that at any root r, if it has odd multiplicity, then the function crosses the x-axis, while if it has
even multiplicity, then it does not cross the x-axis. We note that if 0 = 1, then f (0) > 0 and f (∞) > 0, so
thus we must have an even number of roots (counting multiplicities). Similarly, if 0 = −1, then f (0) < 0 and
f (∞) > 0, so thus we must have an odd number of roots.
We shall prove this by induction on deg f :
Base Case.
n = 1 is fairly obvious. If 0 > 0, then by our Lemma, there are no real roots, and 0 1 = 1. If 0 < 0, then
by our Lemma, there is one real root (as it is odd and at most 1 by the Fundamental Theorem of Algebra),
and 0 1 = −1.
Assume for some positive integer k, the lemma is true for all polynomials f such that deg f ≤ k. We shall
show the statement for deg f = k + 1.
Induction Step.
Now, we have two cases:
Case 1. 0 q = 1, where q is the least positive integer such that q 6= 0.
Then, we have by Rolle’s Theorem (I’m too lazy to prove it),
f 0 (x)
p0 ≥ p − 1 and m0 ≡ p0 (mod 2) and m0 ≥ p0 (by the inductive hypothesis assuming that m0 and p0 have the
same definition with respect to f 0 (x)). By our lemma, we have that considering it with f (x) and f 0 (x),
44
p ≡ p0 (mod 2) (as 0 , q have the same sign), but then we get that
p ≡ p0 ≡ m0 = m (mod 2)
Now, we only need to show that m ≥ p, which is apparent as
p ≤ p0 + 1 ≤ m0 + 1 = m + 1
Case 2. 0 q = −1, where q is the least positive integer such that q 6= 0.
Then, we have by Rolle’s Theorem (I’m too lazy to prove it),
f 0 (x)
p0 ≥ p − 1 and m0 ≡ p0 (mod 2) and m0 ≥ p0 (by the inductive hypothesis assuming that m0 and p0 have the
same definition with respect to f 0 (x)). By our lemma, we have that considering it with f (x) and f 0 (x),
p ≡ 1 + p0 (mod 2) (as 0 , q have the opposite sign), but then we get that
p ≡ p0 + 1 ≡ m0 + 1 = m (mod 2)
Now, we only need to show that m ≥ p, which is apparent as
p ≤ p0 + 1 ≤ m0 + 1 ≤ m
So what about negative roots? We have the following corollary:
Corollary C.6 (Descartes’ Rule of Sign’s Corollary)

The number of negative roots is when we just apply f (−x) to Descartes’ Rule of Signs instead.
Now, this can provide upper bounds on the number of roots. For example, consider the following:
Example C.7
Find the number of nonreal roots of f (x) = x3 + 1.
Solution. By Descartes’ Rule of Signs, we see that there are at most 0 positive roots. By Descartes’ Rule of
Sign’s Corollary, we see there are an odd number of negative roots and at most 1 negative root. Thus there is
exactly 1 negative root, so exactly 1 real root. Thus there are 2 nonreal roots.
Now, there is something that is very useful, and described as the discriminant. We touched on it in our discussion
of AIME I 2011/9 but didn’t actually use it much. We will introduce it here:
Theorem C.8 (Discriminant of a Polynomial)

Define the discriminant of a polynomial is
Y
Disc(p(x)) = a2n−2
n (ri − rj )2
1≤i<j≤n
where the leading coefficient of p(x) is an , the degree is n, and the roots are r1 , r2 , . . . , rn .
45
Note that this immediately implies that the discriminant is 0 if and only if two roots are equal. In addition, all
roots are real if and only if the discriminant is nonnegative (the if part is hard to prove and out of scope - the
only if part is very easy to prove). But how does this help? Well, let’s see an example:
Corollary C.9 (Discrimant of a Quadratic Polynomial)

The discrimant of a quadratic polynomial ax2 + bx + c is
b2 − 4ac
Proof. You may have seen something like this when dealing with the quadratic formula. Let’s talk about it here.
Let the roots be r1 , r2 . By Vieta’s Formulas, we have that
b
σ1 = r1 + r2 = −
a
c
σ2 = r1 r2 =
a
Then, we get that
Disc(p(x)) = a2 (r1 − r2 )2
Let’s expand this:
Disc(p(x)) = a2 (r12 + r22 − 2r1 r2 )
We note that by Newton’s Formulas (or #1 of freeman66’s tactics), we get
Disc(p(x)) = a2 (σ12 − 2σ2 − 2σ2 ) = a2 (σ12 − 4σ2 )
Substituting, we get
Disc(p(x)) = b2 − 4ac
Now, how can we get the discriminant without actually using the roots? We need the following definition:
Definition C.10 (Matrix of Two Polynomials) — Consider two polynomials f (x) and g(x), with degrees m
and n. The resultant is the discriminant of the (m + n) × (m + n) matrix formed by writing the coefficients
of f (x) n times and the coefficients of g(x) m times.
That’s hard to imagine. We need an example:
Example C.11
Find the resultant of the polynomials x2 + 2x + 1 and 2x + 2.
Solution. Well, the resultant is the discriminant of

 
1 2 1
2 2 0 
0 2 2
I know the definition is hard to grasp, but with this example it should be a lot easier. Anyways, we can compute
the resultant to indeed be 0.
How does this relate to anything? The following theorem says it all:
46
Theorem C.12 (Discriminants from Resultants)

The discriminant of the polynomial p(x) is equal to
n
(−1)( 2 )
R(p, p0 )
an
where R(p, p0 ) is the resultant of p(x) and p0 (x), n = deg p, and an is the leading coefficient of p.
This is way too out of scope for me to present. Google it up if you want to see it. However, we can indeed once
again check Discrimant of a Quadratic Polynomial:
Corollary C.13 (Discrimant of a Quadratic Polynomial)

The discrimant of a quadratic polynomial ax2 + bx + c is
b2 − 4ac
Proof. We have that p0 (x) = 2ax + b, so

a b c
−1 1
2a b 0 = − (ab2 + 4a2 c − 2ab2 ) = b2 − 4ac

Disc(p(x)) =
a a
0 2a b
The only other case that would be helpful would be a cubic. I am stating the result, but you can use Discriminants
from Resultants for matrix bashing. Have fun!
Theorem C.14 (Discriminant of a Cubic Polynomial)

The discriminant of a cubic is given by
18abcd − 4b3 d + b2 c2 − 4ac3 − 27a2 d2 .
I can’t really think of a good time this would be useful, except maybe as for what we did for 2005 AIME I
Problem 8 in the beginning of the Appendix.
§D Appendix D: Lagrange Interpolation Formula

This is an addendum to the Polynomial handout, added because it is occasionally useful. Many thanks to
AoPS user anonman for helping us with this!
47
Theorem D.1 (Lagrange Interpolation Formula)

Let a1 , a2 , · · · , an , b1 , b2 , · · · , cn be real numbers. The unique polynomial P of degree < n − 1 such that
P (a1 ) = b1 , P (a2 ) = b2 , · · · , P (an ) = bn
is
n
X (x − a1 ) (x − a2 ) · · · (x − ai−1 ) (x − ai+1 ) · · · (x − an )
P (x) = bi
(ai − a1 ) (ai − a2 ) · · · (ai − ai−1 ) (ai − ai+1 ) · · · (ai − an )
i=1
or more simply,
n
X Y x − am
P (x) = bi
0≤m≤n
ai − am
i=1
m6=i
Example D.2 (USAMO 1975/3)

If P (x) denotes a polynomial of degree n such that
k
P (k) =
k+1
for k = 0, 1, 2, . . . , n, determine P (n + 1).
Solution. It is fairly natural to use Lagrange’s Interpolation Formula on this problem:

n
X k Y n+1−j
P (n + 1) =
k+1 k−j
k=0 j6=k
n (n+1)!
X k n+1−k
= ·
k + 1 k(k − 1)(k − 2) . . . 1 · (−1)(−2) . . . (k − n)
k=0
n
X k (n + 1)!
= (−1)n−k ·
k+1 k!(n + 1 − k)!
k=0
n n
(n + 1)!(−1)n−k
X
n−k n + 1
X
= (−1) −
k (k + 1)!(n + 1 − k)!
k=0 k=0
n+1 n
!
n+1−k n + 1 1 n+2
X X
=− (−1) −1 + · (−1)n+1−k
k n+2 k+1
k=0 k=0
n+1 !
1 X n + 2
=1+ (−1)n+2−(k+1) − (−1)n+2 − 1
n+2 k+1
k=−1
(−1)n + 1
= 1− .
n+2
48
Example D.3 (USAMO 1984/5)

P (x) is a polynomial of degree 3n such that
P (0) = P (3) = · · · = P (3n) = 2,

P (1) = P (4) = · · · = P (3n − 2) = 1,
P (2) = P (5) = · · · = P (3n − 1) = 0, and
P (3n + 1) = 730.
Determine n.
Solution. By Lagrange Interpolation Formula,

   
n n
X Y x−r  X Y x−r 
f (x) = 2  +  ,
3p − r 3p − 2 − r
p=0 0≤r6=3p≤3n p=1 0≤r6=3p−2≤3n
and hence
   
n n
X Y 3n + 1 − r  X  Y 3n + 1 − r 
f (3n + 1) = 2  + .
3p − r 3p − 2 − r
p=0 0≤r6=3p≤3n p=1 0≤r6=3p−2≤3n
After some calculations we get

3n + 1 3n + 1 3n + 1
− . . . 2.(−1)3n − 1 + 1.

f (3n + 1) = − +
0 3 6
Given f (3n + 1) = 730 so we have to find n such that

3n + 1 3n + 1 3n + 1
− . . . 2.(−1)3n − 1 = 729.

− +
0 3 6
Lemma. If p is even, then

π pπ
2p+1 sinp (i)p cos

p p p 3 3
− + − ... = ,
0 3 6 3
and if p is odd,
π pπ
−2p+1 sinp (i)p+1 sin

p p p 3 3
− + − ··· = .
0 3 6 3
(We will not discuss the proof of this lemma in this handout - the idea is Roots of Unity Filter.) Using the
above lemma, we do not get any solutions when n is odd, but when n is even, 3n + 1 = 13 satisfies the required
condition, implying n = 4 is the only solution.
Example D.4 (HMMT Algebra & Number Theory 2020/8)

Let P (x) be the unique polynomial of degree at most 2020 satisfying P (k 2 ) = k for k = 0, 1, 2, . . . , 2020.
Compute P (20212 ).
49
Solution. We know that

P (k 2 ) = k for (k = 0, 1, . . . , 2020).
This gives us the motivation to try Lagrange Interpolation, with points of:
P (0) = 0, P (1) = 1, P (4) = 2, P (9) = 3, . . .
We have ai as all the squares and bi as all the numbers i such that 0 ≤ i ≤ 2020. Now, we will use some fancy
notation to make our work easier. Using Lagrange, we have
2020
X Y x − m2
P (x) = i .
0≤m≤2020
i2 − m2
i=1
m6=i
Y 20212 − m2
Now let us painfully evaluate what P (20212 ) will be. For convenience, let’s first evaluate what
0≤m≤2020
i2 − m2
m6=i
will be.
Y 20212 − m2 Y (2021 + m)(2021 − m)
= .
0≤m≤2020
i2 − m2 0≤m≤2020
(i + m)(i − m)
m6=i m6=i
Let’s see what exactly this expression is saying:

(2021 + 0)(2021 − 0) (2021 + 1)(2021 − 1)
···
(i + 0)(i − 0) (i + 1)(i − 1)
(2021 + i − 1)(2021 − i + 1) (2021 + i + 1)(2021 − i − 1)
···
(i + i − 1)(i − i + 1) (i + i + 1)(i − i − 1)
(2021 + 2019)(2021 − 2019) (2021 + 2020)(2021 − 2020)
.
(i + 2019)(i − 2019) (i + 2020)(i − 2020)
Now, let us take the product of all numerators and then divide it by the product of all denominators:
• Numerators: If you inspect closely, you will notice that we have a 4041!, an extra 2021, and we have to
exclude 2021 + i and 2021 − i from the factorial as m 6= i. Hence, the product is
4041! · 2021
.
(2021 + i)(2021 − i)
• Denominators: We have a (2021 + i)!, a (2020 − i)!, an extra i, and then we need to exclude 2i from
(2021 + i)! since m 6= i. Also notice that the sign of (2020 − i)! depends on parity of i. Hence, the product is
(2020 + i)!(2020 − i)!
(−1)i .
2
Let’s finish this monster off:
4041!·2021
Y (2021 + m)(2021 − m) (2021+i)(2021−i)
=
0≤m≤2020
(i + m)(i − m) (−1)i (2020+i)!(2020−i)!
2
m6=i
(−1)i · 4042!
=
(2021 + i)!(2021 − i)!

i 4042
= (−1) .
2021 − i
Sweet! But before we truly destroy the beast, we use Pascal’s Identity.
50
Theorem D.5 (Pascal’s identity)

For any positive integer n and k, Pascal’s identity states that

n n−1 n−1
= +
k k−1 k
Let’s begin the final battle:

2020
X Y x − m2
P (x) = i
0≤m≤2020
i2 − m2
i=1
m6=i
2020
X Y 20212 − m2
= i
0≤m≤2020
i2 − m2
i=1
m6=i
2020
X
i 4042
= i(−1)
2021 − i
i=1
2021
!
X 4041 4041
= 2021 − (−1)i+1 i +
2021 − i 2020 − i
i=0
2021 !
X 4041
= 2021 − (−1)i+1
2021 − i
i=1
2021 !
X 4040 4040
= 2021 − (−1)i+1 +
2021 − i 2020 − i
i=1

4040
= 2021 − .
2020
The beast has been conquered.
Exercise D.6 (USAMO 2002/3). Prove that any monic polynomial (a polynomial with leading coefficient
1) of degree n with real coefficients is the average of two monic polynomials of degree n with n real roots.
Exercise D.7 (AIME I 2015/10). Let f (x) be a third-degree polynomial with real coefficients satisfying
|f (1)| = |f (2)| = |f (3)| = |f (5)| = |f (6)| = |f (7)| = 12.
Find |f (0)|.
Exercise D.8 (HMMT February Algebra & Number Theory 2017/6). A polynomial P of degree 2015
satisfies the equation P (n) = m12 for n = 1, 2, . . . , 2016. Find b2017P (2017)c.
§E Appendix E: List of Theorems, Corollaries, and Definitions
List of Theorems
51
2.1 Theorem - Fundamental Theorem of Algebra 5
2.8 Theorem - Unique Factorization of Polynomials 5
2.10 Theorem - Factor Theorem 6
2.11 Theorem - Remainder Theorem 6
2.19 Theorem - Rational Root Theorem 8
3.7 Theorem - Vieta’s Formulas 11
3.9 Theorem - Binomial Theorem 11
4.3 Theorem - Fundamental Theorem of Symmetric Polynomials 13
4.5 Theorem - Newton’s Formulas 13
5.13 Theorem - Complex Conjugate Theorem 18
5.18 Theorem - De Moivre’s Theorem 20
5.19 Theorem - Euler’s Formula 20
5.21 Theorem - De Moivre’s Theorem 20
6.1 Theorem - Difference of Polynomials 23
A.1 Theorem - Fundamental Theorem of Algebra 39
A.2 Theorem - Liouville’s Theorem 39
A.3 Theorem - Fundamental Theorem of Symmetric Polynomials 39
C.1 Theorem - Intermediate Value Theorem 42
C.3 Theorem - Descartes’ Rule of Signs 44
C.8 Theorem - Discriminant of a Polynomial 45
C.12 Theorem - Discriminants from Resultants 47
C.14 Theorem - Discriminant of a Cubic Polynomial 47
D.1 Theorem - Lagrange Interpolation Formula 48
D.5 Theorem - Pascal’s identity 51
List of Corollaries
2.5 Corollary - Number of Roots Corollary 5
2.7 Corollary - Zero Polynomial Corollary 5
52
C.6 Corollary - Descartes’ Rule of Sign’s Corollary 45
C.9 Corollary - Discrimant of a Quadratic Polynomial 46
C.13 Corollary - Discrimant of a Quadratic Polynomial 47
List of Definitions
4.1 Definition - Elementary Symmetric Polynomial 13
4.2 Definition - k-Variable Symmetric Polynomial 13
5.1 Definition - Complex Number 16
5.2 Definition - Conjugate 16
5.9 Definition - Modulus/Magnitude 17
5.24 Definition - Roots of Unity 21
C.10 Definition - Matrix of Two Polynomials 46
53

Polynomials in The AIME

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Polynomials in The AIME

Uploaded by

Copyright:

Available Formats

AIME Handout

Polynomials in the AIME

There will be a lot of this in the handout.

I had a polynomial once. My doctor removed it. - Michael Grant, “Gone”

6 A Small Bit Of Number Theory 23

7 Advanced Algebraic Manipulations 24

8 Worked Through Problems 28

9 Parting Words and Final Problems 32

A Appendix A: Proof of Results 39

B Appendix B: Polynomial Division 40

C Appendix C: Real Roots 42

D Appendix D: Lagrange Interpolation Formula 47

E Appendix E: List of Theorems, Corollaries, and Definitions 51

Evan Chen’s Personal Sty File

Me! Say hi!

And Evan says he would like this here for evan.sty:

Boost Software License - Version 1.0 - August 17th, 2003

He also helped us with the hint formatting. Evan is a LATEXgod!

Problem 1 (2016 AIME I Problem 11)

Problem 2 (1984 AIME Problem 15)

Theorem 2.1 (Fundamental Theorem of Algebra)

Corollary 2.5 (Number of Roots Corollary)

This leads to the following corollary:

Corollary 2.7 (Zero Polynomial Corollary)

Theorem 2.8 (Unique Factorization of Polynomials)

Theorem 2.10 (Factor Theorem)

This leads to the following generalization (proved in Appendix B: Polynomial Division):

Theorem 2.11 (Remainder Theorem)

Example 2.12 (AMC 12A 2017/23)

f (x) = x4 + x3 + bx2 + 100x + c.

x4 + x3 + bx2 + 100x + c = (x3 + ax2 + x + 10)(x + r)

Thus, we get that

We know ar + 1...but isn’t that a bit big?

and similarly c = 10r = 900. Thus, we can plug it in to get

f (1) = 1 + 1 − 8009 + 100 + 900 = −7007

Theorem 2.19 (Rational Root Theorem)

Example 2.20 (naman12)

Solution. [Walkthrough] I’ll only explain a walkthrough on how to solve this:

(b) Write all factors of our constant term.

(d) Consider all roots. Try to find them.

Example 2.21 (AIME I 2011/9)

Let’s cap it off with a few exercises:

In this problem, taking a = 8, b = 3, and c = 1. 32 = 9 ≤ 32 = 4 · 8, so there are no real roots.

Isn’t this complicated?

Exercise 3.6. Can you generalize what σn is for n variables? Hints: 61

Theorem 3.7 (Vieta’s Formulas)

Example 3.8 (AIME I 2001/3)

Theorem 3.9 (Binomial Theorem)

Equipped with this, we get that

Exercise 3.11. Prove Binomial Theorem. Hints: 13

Let’s see another example:

Example 3.13 (AIME 1996/5)

t = −(a + b)(b + c)(c + a) = −(2abc + a2 b + ab2 + b2 c + bc2 + c2 a + ca2 )

t = −(k − a)(k − b)(k − c)

8x3 + 1001x + 2008 = 0.

Find (r + s)3 + (s + t)3 + (t + r)3 . Hints: 17

Definition 4.1 (Elementary Symmetric Polynomial) — An elementary symmetric polynomial is any

Furthermore, we have the following definition

Definition 4.2 (k-Variable Symmetric Polynomial) — A symmetric polynomial in k variables is basically

Now, this leads to the following powerful theorem:

Theorem 4.3 (Fundamental Theorem of Symmetric Polynomials)

For example, try this exercise:

Theorem 4.5 (Newton’s Formulas)